def parse_index_file(self, fname, base_url=URL()):
parser = SiteParser()
startpage_rule = ParserRule()
startpage_rule.add_activate_rule_level([('div', 'class', 'thumbs200'),
('div', 'class', 'thumbs300')])
startpage_rule.add_process_rule_level('a', {'href'})
startpage_rule.add_process_rule_level('img', {'src', 'alt'})
startpage_rule.set_attribute_modifier_function('href', self.get_href)
parser.add_rule(startpage_rule)
startpage_pages_rule = ParserRule()
startpage_pages_rule.add_activate_rule_level([('div', 'class', 'menu')
])
startpage_pages_rule.add_process_rule_level('a', {'href'})
startpage_pages_rule.set_attribute_modifier_function(
'href', lambda x: base_url.domain() + x)
startpage_pages_rule.set_attribute_filter_function(
'href', lambda txt: '/st/' in txt)
parser.add_rule(startpage_pages_rule)
picture_rule = ParserRule()
picture_rule.add_activate_rule_level([('div', 'class',
'gallery-thumbs')])
picture_rule.add_process_rule_level('a', set())
picture_rule.add_process_rule_level('img', {'src'})
picture_rule.set_attribute_modifier_function(
'src', self.process_picture_address)
parser.add_rule(picture_rule)
for s in open(fname):
parser.feed(s)
result = ParseResult()
if len(picture_rule.get_result()) > 0:
result.set_type('pictures')
for f in picture_rule.get_result():
result.add_full(
FullPictureInfo(abs_href=URL(f['src']),
rel_name=f['src'].rpartition('/')[2]))
return result
if len(startpage_rule.get_result()) > 0:
# print('Startpage rule')
result.set_type('hrefs')
for item in startpage_rule.get_result():
result.add_thumb(
ThumbInfo(thumb_url=URL(item['src']),
href=URL(item['href']),
popup=item.get('alt', '')))
for item in startpage_pages_rule.get_result(['href', 'data']):
result.add_page(ControlInfo(item['data'], URL(item['href'])))
#
# for item in startpage_hrefs_rule.get_result(['href', 'data']):
# result.add_control(ControlInfo(item['data'], URL(item['href'])))
return result
def parse_index_file(self, fname, base_url=URL()):
parser = SiteParser()
startpage_rule = ParserRule()
startpage_rule.add_activate_rule_level([('div', 'class', 'video')])
startpage_rule.add_process_rule_level('a', {'href', 'title'})
startpage_rule.add_process_rule_level('img', {'src', 'alt'})
startpage_rule.set_attribute_modifier_function(
'href', lambda x: self.get_href(x, base_url))
startpage_rule.set_attribute_modifier_function(
'src', lambda x: self.get_href(x, base_url))
parser.add_rule(startpage_rule)
channels_rule = ParserRule()
channels_rule.add_activate_rule_level([('ul', 'class', 'channels')])
channels_rule.add_process_rule_level('a', {'href', 'title'})
channels_rule.add_process_rule_level('div', {})
channels_rule.add_process_rule_level('img', {'src', 'alt'})
channels_rule.set_attribute_modifier_function(
'href', lambda x: self.get_href(x, base_url).replace('*', '/'))
channels_rule.set_attribute_modifier_function(
'src', lambda x: self.get_href(x, base_url))
parser.add_rule(channels_rule)
startpage_pages_rule = ParserRule()
startpage_pages_rule.add_activate_rule_level([
('ul', 'class', 'pagination pagination-lg')
])
# startpage_pages_rule.add_activate_rule_level([('a', 'class', 'current')])
startpage_pages_rule.add_process_rule_level('a', {'href'})
startpage_pages_rule.set_attribute_modifier_function(
'href', lambda x: self.get_href(x, base_url))
parser.add_rule(startpage_pages_rule)
startpage_hrefs_rule = ParserRule()
startpage_hrefs_rule.add_activate_rule_level([
('ul', 'class', 'nav nav-stacked navigation')
])
# startpage_hrefs_rule.add_activate_rule_level([('a', 'class', 'current')])
startpage_hrefs_rule.add_process_rule_level('a', {'href'})
# startpage_hrefs_rule.set_attribute_filter_function('href',lambda x: '/videos/' in x)
startpage_hrefs_rule.set_attribute_modifier_function(
'href', lambda x: self.get_href(x, base_url))
parser.add_rule(startpage_hrefs_rule)
#
video_rule = ParserRule()
video_rule.add_activate_rule_level([('div', 'id', 'player')])
video_rule.add_process_rule_level('script', {})
video_rule.set_attribute_filter_function(
'data', lambda text: 'jwplayer' in text)
# video_rule.set_attribute_modifier_function('src',lambda x:self.get_href(x,base_url))
parser.add_rule(video_rule)
video2_rule = ParserRule()
video2_rule.add_activate_rule_level([('div', 'id', 'video')])
video2_rule.add_process_rule_level('script', {'src'})
video2_rule.set_attribute_filter_function(
'src', lambda text: 'pornbraze.com/' in text)
video2_rule.set_attribute_modifier_function(
'src', lambda x: self.get_href(x, base_url))
parser.add_rule(video2_rule)
#
gallery_href_rule = ParserRule()
gallery_href_rule.add_activate_rule_level([
('div', 'class', 'col-xs-12 col-sm-12 col-md-12')
])
gallery_href_rule.add_process_rule_level('a', {'href'})
gallery_href_rule.set_attribute_modifier_function(
'href', lambda x: self.get_href(x, base_url))
parser.add_rule(gallery_href_rule)
self.proceed_parcing(parser, fname)
result = ParseResult()
if video_rule.is_result(): # len(video_rule.get_result()) > 0:
urls = list()
for item in video_rule.get_result():
# print(item['data'])
script = item['data'].replace(' ', '')
if 'sources:[{' in script:
txt = '[{' + self.quotes(item['data'].replace(' ', ''),
'sources:[{', '}]') + '}]'
j = json.loads(txt)
for j_data in j:
# print(j_data)
if j_data['file'] is not '':
data = dict(text=j_data['label'],
url=URL(j_data['file'] + '*'))
urls.append(data)
elif 'sources:' in script:
if video2_rule.is_result(['src']):
# print(video2_rule.get_result())
php_url = URL(
video2_rule.get_result(['src'])[0]['src'])
# print(php_url)
res = load(php_url)
# print(res.text)
bitrates = self.quotes(res.text, "'bitrates':[{",
"}]").split('},{')
# print(bitrates)
for line in bitrates:
print(line)
video_url = self.quotes(line, "'file':'", "'")
label = self.quotes(line, 'label:"', '"')
data = dict(text=label, url=URL(video_url + '*'))
urls.append(data)
if len(urls) == 1:
video = MediaData(urls[0]['url'])
elif len(urls) > 1:
video = MediaData(urls[0]['url'])
for item in urls:
video.add_alternate(item)
else:
return result
result.set_type('video')
result.set_video(video)
for f in gallery_href_rule.get_result(['data', 'href']):
# print(f)
href = f['href'].replace('*', '/')
label = f['data']
if '/users/' in href:
href = href + '/videos/public/'
label = '"' + label + '"'
result.add_control(ControlInfo(label, URL(href)))
return result
if startpage_rule.is_result() or channels_rule.is_result():
result.set_type('hrefs')
for item in startpage_rule.get_result(['href']):
result.add_thumb(
ThumbInfo(thumb_url=URL(item['src']),
href=URL(item['href']),
popup=item.get('alt', item.get('title', ''))))
for item in channels_rule.get_result(['href']):
result.add_thumb(
ThumbInfo(thumb_url=URL(item['src']),
href=URL(item['href']),
popup=item.get('alt', item.get('title', ''))))
for item in startpage_pages_rule.get_result(['href', 'data']):
result.add_page(ControlInfo(item['data'], URL(item['href'])))
for item in startpage_hrefs_rule.get_result():
label = item['href'].strip('*/').rpartition('/')[2]
result.add_control(ControlInfo(label, URL(item['href'])))
return result
def parse_index_file(self, fname, base_url=URL()):
parser = SiteParser()
startpage_rule = ParserRule()
startpage_rule.add_activate_rule_level([('div', 'class', 'thumbs'),
('div', 'class', 'content_box domain2'),
('div', 'class', 'video_list'),
('div', 'class', 'video_list_models'),
('div', 'class', 'pics_list'),
('div', 'class', 'movie_thumbs')])
startpage_rule.add_process_rule_level('a', {'href'})
startpage_rule.add_process_rule_level('img', {'src', 'alt'})
startpage_rule.set_attribute_modifier_function('href', self.get_href)
parser.add_rule(startpage_rule)
startpage_pages_rule = ParserRule()
startpage_pages_rule.add_activate_rule_level([('div', 'class', 'pages'),
('div', 'class', 'pg')])
startpage_pages_rule.add_process_rule_level('a', {'href'})
startpage_pages_rule.set_attribute_modifier_function('href', lambda x: base_url.domain() + x)
parser.add_rule(startpage_pages_rule)
startpage_hrefs_rule = ParserRule()
startpage_hrefs_rule.add_activate_rule_level([('li', 'class', 'orange dropdown')])
startpage_hrefs_rule.add_process_rule_level('a', {'href'})
startpage_hrefs_rule.set_attribute_filter_function('href', lambda txt: '/st/' in txt)
parser.add_rule(startpage_hrefs_rule)
video_rule = ParserRule()
video_rule.add_activate_rule_level([('div', 'class', 'player')])
video_rule.add_process_rule_level('script', {})
video_rule.set_attribute_filter_function('data', lambda text: 'video_url:' in text)
parser.add_rule(video_rule)
picture_rule = ParserRule()
picture_rule.add_activate_rule_level([('div', 'class', 'thumb_box')])
picture_rule.add_process_rule_level('a', set())
picture_rule.add_process_rule_level('img', {'src'})
picture_rule.set_attribute_modifier_function('src', lambda text: text.replace('t.jpg', '.jpg'))
parser.add_rule(picture_rule)
picture_href_rule = ParserRule()
picture_href_rule.add_activate_rule_level([('div', 'class', 'crumbles'),
('div', 'class', 'tags')])
picture_href_rule.add_process_rule_level('a', {'href', 'title'})
parser.add_rule(picture_href_rule)
for s in open(fname):
parser.feed(s)
result = ParseResult()
if video_rule.is_result():
video = MediaData(URL(self.get_attr_from_script(video_rule.get_result()[0]['data'])))
result.set_video(video)
result.set_type('video')
for f in picture_href_rule.get_result():
# print(f)
result.add_control(ControlInfo(f['data'], URL(f['href'])))
return result
if len(startpage_rule.get_result()) > 0:
# print('Startpage rule')
result.set_type('hrefs')
for item in startpage_rule.get_result():
result.add_thumb(
ThumbInfo(thumb_url=URL(item['src']), href=URL(item['href']), popup=item.get('alt', '')))
for item in startpage_pages_rule.get_result(['href', 'data']):
result.add_page(ControlInfo(item['data'], URL(item['href'])))
for item in startpage_hrefs_rule.get_result(['href', 'data']):
result.add_control(ControlInfo(item['data'], URL(item['href'])))
if len(picture_rule.get_result()) > 0:
result.set_type('pictures')
for f in picture_rule.get_result():
result.add_full(FullPictureInfo(rel_name=f['src']))
for f in picture_href_rule.get_result():
# print(f)
result.add_control(ControlInfo(f['title'], URL(f['href'])))
return result
def parse_index_file(self, fname, base_url=URL()):
parser = SiteParser()
# def star_get_url(txt=''):
# return txt.partition('(')[2].partition(')')[0]
startpage_rule = ParserRule()
startpage_rule.add_activate_rule_level([
('div', 'class', 'list_videos'), ('div', 'class', 'list_albums'),
('div', 'class', 'list_videos model-girls-list'),
('div', 'class', 'list_videos list_channel')
])
startpage_rule.add_process_rule_level('a', {'href'})
startpage_rule.add_process_rule_level('img', {'src', 'alt'})
startpage_rule.set_attribute_modifier_function(
'href', lambda x: self.get_href(x, base_url))
startpage_rule.set_attribute_modifier_function(
'src', lambda x: self.get_href(x, base_url))
parser.add_rule(startpage_rule)
startpage_pages_rule = ParserRule()
startpage_pages_rule.add_activate_rule_level([('div', 'class',
'pagination')])
# startpage_pages_rule.add_activate_rule_level([('a', 'class', 'current')])
startpage_pages_rule.add_process_rule_level('a', {'href'})
startpage_pages_rule.set_attribute_modifier_function(
'href', lambda x: self.get_href(x, base_url))
parser.add_rule(startpage_pages_rule)
startpage_hrefs_rule = ParserRule()
startpage_hrefs_rule.add_activate_rule_level([('div', 'class',
'model-alpha')])
# startpage_hrefs_rule.add_activate_rule_level([('a', 'class', 'current')])
startpage_hrefs_rule.add_process_rule_level('a', {'href'})
# startpage_hrefs_rule.set_attribute_filter_function('href',lambda x: '/videos/' in x)
startpage_hrefs_rule.set_attribute_modifier_function(
'href', lambda x: self.get_href(x, base_url))
parser.add_rule(startpage_hrefs_rule)
#
video_rule = ParserRule()
video_rule.add_activate_rule_level([('div', 'class', 'vids')])
video_rule.add_process_rule_level('script', {})
video_rule.set_attribute_filter_function(
'data', lambda text: 'video_url' in text)
parser.add_rule(video_rule)
#
gallery_href_rule = ParserRule()
gallery_href_rule.add_activate_rule_level([('div', 'class',
'video-categories')])
gallery_href_rule.add_process_rule_level('a', {'href'})
gallery_href_rule.set_attribute_modifier_function(
'href', lambda x: self.get_href(x, base_url))
parser.add_rule(gallery_href_rule)
gallery_user_rule = ParserRule(collect_data=True)
gallery_user_rule.add_activate_rule_level([('div', 'class',
'video-added-info')])
gallery_user_rule.add_process_rule_level('a', {'href'})
# gallery_user_rule.set_attribute_modifier_function('href', lambda x: self.get_href(x+'/videos',base_url))
gallery_user_rule.set_attribute_filter_function(
'href', lambda x: '/members/' in x)
parser.add_rule(gallery_user_rule)
photo_rule = ParserRule()
photo_rule.add_activate_rule_level([('div', 'class', 'zoom-gallery')])
photo_rule.add_process_rule_level('a', {'href'})
# photo_rule.set_attribute_filter_function('href', lambda text: '/photos/' in text)
photo_rule.set_attribute_modifier_function(
'href', lambda x: self.get_href(x, base_url))
parser.add_rule(photo_rule)
self.proceed_parcing(parser, fname)
result = ParseResult()
def add_href_and_user_to_result():
if gallery_user_rule.is_result(['href']):
for item in gallery_user_rule.get_result(['href']):
# print(item)
username = item['data'].strip().partition(
'Added by ')[2].partition(' ')[0]
# print(username)
if username != '':
result.add_control(
ControlInfo('"' + username + ' videos"',
URL(item['href'] + 'public_videos/')))
result.add_control(
ControlInfo('"' + username + ' photos"',
URL(item['href'] + 'albums/')))
for f in gallery_href_rule.get_result(['data', 'href']):
result.add_control(ControlInfo(f['data'], URL(f['href'])))
if video_rule.is_result(): # len(video_rule.get_result()) > 0:
# for item in video_rule.get_result():
# print('=============================')
# print(item['data'])
script = video_rule.get_result()[0]['data'].replace(' ', '')
# print(script)
url = script.partition("video_url:'")[2].partition("'")[0]
# print(url)
video = MediaData(URL(url))
result.set_type('video')
result.set_video(video)
add_href_and_user_to_result()
return result
if photo_rule.is_result():
result.set_type('pictures')
base_dir = base_url.get_path(base=Setting.base_dir) + base_url.get(
).rpartition('/')[2] + '/'
result.set_gallery_path(base_dir)
# print(base_dir)
for item in photo_rule.get_result():
name = item['href'].rpartition('/')[2].strip('*')
picture = FullPictureInfo(abs_href=URL(item['href']),
rel_name=name)
picture.set_base(base_dir)
result.add_full(picture)
add_href_and_user_to_result()
return result
if startpage_rule.is_result(): # len(startpage_rule.get_result()) > 0:
result.set_type('hrefs')
for item in startpage_rule.get_result(['href']):
# print(item)
result.add_thumb(
ThumbInfo(thumb_url=URL(item['src']),
href=URL(item['href']),
popup=item.get('alt', '')))
for item in startpage_pages_rule.get_result(['href', 'data']):
result.add_page(ControlInfo(item['data'], URL(item['href'])))
if len(startpage_hrefs_rule.get_result(['href'])) > 0:
for item in startpage_hrefs_rule.get_result(['href', 'data']):
result.add_control(
ControlInfo(item.get('data', item.get('title', '')),
URL(item['href'])))
return result
def parse_index_file(self, fname, base_url=URL()):
parser = SiteParser()
startpage_rule = ParserRule()
startpage_rule.add_activate_rule_level([
('span', 'class', 'thumb_container_box short'),
('span', 'class', 'thumb_container_box long')
])
startpage_rule.add_process_rule_level('a', {'href', 'title'})
startpage_rule.add_process_rule_level('img', {'src'})
startpage_rule.set_attribute_modifier_function(
'href', lambda x: self.get_href(x, base_url))
parser.add_rule(startpage_rule)
startpage_pages_rule = ParserRule()
startpage_pages_rule.add_activate_rule_level([('div', 'class',
'main-sectionpaging')])
# startpage_pages_rule.add_activate_rule_level([('a', 'class', 'current')])
startpage_pages_rule.add_process_rule_level('a', {'href'})
startpage_pages_rule.set_attribute_modifier_function(
'href', lambda x: util.get_href(x, base_url))
parser.add_rule(startpage_pages_rule)
startpage_hrefs_rule = ParserRule()
startpage_hrefs_rule.add_activate_rule_level([
('ul', 'class', 'simple-list simple-list--channels')
])
# startpage_hrefs_rule.add_activate_rule_level([('a', 'class', 'current')])
startpage_hrefs_rule.add_process_rule_level('a', {'href', 'title'})
# startpage_hrefs_rule.set_attribute_modifier_function('href', lambda x: self.get_href(x,base_url) + '*')
parser.add_rule(startpage_hrefs_rule)
#
video_rule = ParserRule()
video_rule.add_activate_rule_level([('video', '', '')])
video_rule.add_process_rule_level('source', {'src', 'id'})
# video_rule.set_attribute_filter_function('data', lambda text: 'jwplayer' in text)
video_rule.set_attribute_modifier_function('src',
lambda txt: txt + '*')
parser.add_rule(video_rule)
video_script_rule = ParserRule()
video_script_rule.add_activate_rule_level([('body', '', '')])
video_script_rule.add_process_rule_level('script', {})
video_script_rule.set_attribute_filter_function(
'data', lambda text: 'shows:' in text)
# video_script_rule.set_attribute_modifier_function('src',lambda txt:txt+'*')
parser.add_rule(video_script_rule)
gallery_rule = ParserRule()
gallery_rule.add_activate_rule_level([('div', 'id', 'slideshow')])
gallery_rule.add_process_rule_level('a', {'index'})
gallery_rule.add_process_rule_level('img', {'src'})
# video_rule.set_attribute_filter_function('data', lambda text: 'jwplayer' in text)
gallery_rule.set_attribute_modifier_function(
'src', lambda txt: txt.replace('/thumbs/', '/'))
parser.add_rule(gallery_rule)
#
gallery_href_rule = ParserRule()
gallery_href_rule.add_activate_rule_level([('div', 'class', 'added')])
gallery_href_rule.add_process_rule_level('a', {'href'})
gallery_href_rule.set_attribute_modifier_function(
'href', lambda x: util.get_href(x, base_url))
parser.add_rule(gallery_href_rule)
self.proceed_parcing(parser, fname)
result = ParseResult()
if video_script_rule.is_result() or video_rule.is_result():
files = set()
default_vid = None
for item in video_script_rule.get_result():
script = item['data'].replace(' ', '')
streams = up.unquote(self.quotes(script, '"streams":[', ']'))
while '"file":"' in streams:
split = streams.partition('"file":"')[2].partition('"')
streams = split[2]
files.add(split[0] + '*')
for item in video_rule.get_result():
files.add(item['src'])
if 'id' not in item:
default_vid = item['src']
if len(files) == 0:
return result
if default_vid is None:
default_vid = files.pop()
else:
files.discard(default_vid)
urls = UrlList()
urls.add('Default', URL(default_vid))
for item in files:
urls.add(item[-5:-1], URL(item))
result.set_video(urls.get_media_data())
for f in gallery_href_rule.get_result(['href']):
label = f['data'].strip()
result.add_control(ControlInfo(label, URL(f['href'])))
return result
if gallery_rule.is_result():
result.set_type('pictures')
url = URL(gallery_rule.get_result()[0]['src'] + '*')
base_dir = url.get_path(base=Setting.base_dir)
result.set_gallery_path(base_dir)
for f in gallery_rule.get_result():
fname = int(f['index'])
# print(fname)
picture = FullPictureInfo(
abs_href=URL(f['src'] + '*'),
rel_name='pic{0:03d}.jpg'.format(fname))
picture.set_base(base_dir)
result.add_full(picture)
for f in gallery_href_rule.get_result(['href']):
label = f['data'].strip()
result.add_control(ControlInfo(label, URL(f['href'])))
return result
if startpage_rule.is_result(): # len(startpage_rule.get_result()) > 0:
for item in startpage_rule.get_result(['href']):
result.add_thumb(
ThumbInfo(thumb_url=URL(item['src']),
href=URL(item['href']),
popup=item.get('title', '')))
for item in startpage_pages_rule.get_result(['href', 'data']):
result.add_page(ControlInfo(item['data'], URL(item['href'])))
if len(startpage_hrefs_rule.get_result(['href'])) > 0:
for item in startpage_hrefs_rule.get_result(['href', 'data']):
result.add_control(
ControlInfo(item.get('title', item.get('data', '')),
URL(item['href'])))
return result
def parse_index_file(self, fname, base_url=URL()):
parser = SiteParser()
# def star_get_url(txt=''):
# return txt.partition('(')[2].partition(')')[0]
startpage_rule = ParserRule()
startpage_rule.add_activate_rule_level([('div', 'class', 'well wellov well-sm'),
('div', 'class', 'col-sm-4 m-t-15')])
startpage_rule.add_process_rule_level('a', {'href'})
startpage_rule.add_process_rule_level('img', {'src', 'alt', 'data-original'})
startpage_rule.set_attribute_modifier_function('href', lambda x: self.get_href(x, base_url))
parser.add_rule(startpage_rule)
startpage_pages_rule = ParserRule()
startpage_pages_rule.add_activate_rule_level([('ul', 'class', 'pagination')])
# startpage_pages_rule.add_activate_rule_level([('a', 'class', 'current')])
startpage_pages_rule.add_process_rule_level('a', {'href'})
startpage_pages_rule.set_attribute_modifier_function('href', lambda x: self.get_href(x, base_url))
parser.add_rule(startpage_pages_rule)
startpage_hrefs_rule = ParserRule()
startpage_hrefs_rule.add_activate_rule_level([('div', 'class', 'btn-group')])
# startpage_hrefs_rule.add_activate_rule_level([('a', 'class', 'current')])
startpage_hrefs_rule.add_process_rule_level('a', {'href'})
# startpage_hrefs_rule.set_attribute_filter_function('href',lambda x: '/videos/' in x)
startpage_hrefs_rule.set_attribute_modifier_function('href', lambda x: self.get_href(x, base_url))
parser.add_rule(startpage_hrefs_rule)
#
video_rule = ParserRule()
video_rule.add_activate_rule_level([('div', 'class', 'container')])
video_rule.add_process_rule_level('script', {})
video_rule.set_attribute_filter_function('data', lambda text: 'nuevoplayer' in text)
parser.add_rule(video_rule)
#
gallery_href_rule = ParserRule()
gallery_href_rule.add_activate_rule_level([('div', 'class', 'm-t-10 overflow-hidden catmenu')])
gallery_href_rule.add_process_rule_level('a', {'href'})
gallery_href_rule.set_attribute_modifier_function('href', lambda x: self.get_href(x, base_url))
parser.add_rule(gallery_href_rule)
gallery_user_rule = ParserRule()
gallery_user_rule.add_activate_rule_level([('div', 'class', 'pull-left user-container')])
gallery_user_rule.add_process_rule_level('a', {'href'})
gallery_user_rule.set_attribute_modifier_function('href', lambda x: self.get_href(x + '/videos', base_url))
gallery_user_rule.set_attribute_filter_function('href', lambda x: '/user/' in x)
parser.add_rule(gallery_user_rule)
photo_rule = ParserRule()
photo_rule.add_activate_rule_level([('div', 'class', 'panel-body')])
photo_rule.add_process_rule_level('a', {'href'})
photo_rule.set_attribute_filter_function('href', lambda text: '/photos/' in text)
photo_rule.set_attribute_modifier_function('href', lambda x: self.get_href(x, base_url))
parser.add_rule(photo_rule)
self.proceed_parcing(parser, fname)
result = ParseResult()
if video_rule.is_result(): # len(video_rule.get_result()) > 0:
script = video_rule.get_result()[0]['data']
urls = list()
for item in script.split(';'):
if 'var video' in item:
part = item.strip().partition('video_')[2].partition('=')
data = dict(text=part[0], url=URL(part[2].strip("'") + '*'))
urls.append(data)
if len(urls) == 1:
video = MediaData(urls[0]['url'])
elif len(urls) > 1:
video = MediaData(urls[-1]['url'])
for item in urls:
video.add_alternate(item)
else:
return result
result.set_type('video')
result.set_video(video)
#
# for f in gallery_channel_rule.get_result(['data', 'href']):
# result.add_control(ControlInfo(f['data'], URL(f['href'])))
if gallery_user_rule.is_result():
for item in gallery_user_rule.get_result():
# print(item)
# print('*'+item['data'].strip()+'*')
username = item['data'].strip()
if username != '':
result.add_control(ControlInfo('"' + username + '"', URL(item['href'])))
for f in gallery_href_rule.get_result(['data', 'href']):
result.add_control(ControlInfo(f['data'], URL(f['href'])))
return result
if photo_rule.is_result():
result.set_type('pictures')
base_dir = base_url.get_path(base=Setting.base_dir)
result.set_gallery_path(base_dir)
for item in photo_rule.get_result():
name = item['href'].rpartition('/')[2].strip('*')
picture = FullPictureInfo(abs_href=URL(item['href']), rel_name=name)
picture.set_base(base_dir)
result.add_full(picture)
return result
if startpage_rule.is_result(): # len(startpage_rule.get_result()) > 0:
result.set_type('hrefs')
for item in startpage_rule.get_result(['href']):
# print(item)
t_url = self.get_href(item.get('data-original', item['src']), base_url)
t_href = item['href']
# print(t_href)
if '/album/' in t_href:
# print(t_href)
t_href = t_href.replace('/album/', '/album/slideshow/')
# print(t_href)
result.add_thumb(ThumbInfo(thumb_url=URL(t_url), href=URL(t_href), popup=item.get('alt', '')))
for item in startpage_pages_rule.get_result(['href', 'data']):
result.add_page(ControlInfo(item['data'], URL(item['href'])))
if len(startpage_hrefs_rule.get_result(['href'])) > 0:
for item in startpage_hrefs_rule.get_result(['href', 'data']):
result.add_control(ControlInfo(item.get('title', item.get('data', '')), URL(item['href'])))
return result
def parse_index_file(self, fname, base_url=URL()):
parser = SiteParser()
startpage_rule = ParserRule()
startpage_rule.add_activate_rule_level([('ul', 'class', 'thumbs'),
('li', 'class', 'category')])
startpage_rule.add_process_rule_level('a', {'href'})
startpage_rule.add_process_rule_level('img',
{'src', 'alt', 'data-original'})
startpage_rule.set_attribute_modifier_function(
'href', lambda x: self.get_href(x, base_url))
parser.add_rule(startpage_rule)
startpage_pages_rule = ParserRule()
startpage_pages_rule.add_activate_rule_level([('div', 'class',
'pager paging')])
# startpage_pages_rule.add_activate_rule_level([('a', 'class', 'current')])
startpage_pages_rule.add_process_rule_level(
'span', {'data-query-key', 'data-query-value'})
# startpage_pages_rule.set_attribute_modifier_function('href', lambda x: self.get_href(x,base_url))
parser.add_rule(startpage_pages_rule)
channels_rule = ParserRule()
channels_rule.add_activate_rule_level([('ul', 'class', 'tag-150-list')
])
channels_rule.add_process_rule_level('a', {'href'})
channels_rule.add_process_rule_level('img', {'src'})
channels_rule.set_attribute_filter_function(
'href', lambda x: '/channel/' in x or '/prime/' in x)
channels_rule.set_attribute_modifier_function(
'href', lambda x: self.get_href(x, base_url))
parser.add_rule(channels_rule)
channel_categories_rule = ParserRule()
channel_categories_rule.add_activate_rule_level([
('ul', 'class', 'link-tag-list long-col')
])
# startpage_pages_rule.add_activate_rule_level([('a', 'class', 'current')])
channel_categories_rule.add_process_rule_level(
'span', {'data-query-key', 'data-query-value'})
# startpage_pages_rule.set_attribute_modifier_function('href', lambda x: self.get_href(x,base_url))
parser.add_rule(channel_categories_rule)
video_rule = ParserRule()
video_rule.add_activate_rule_level([('div', 'class',
'player-container')])
video_rule.add_process_rule_level('script', {})
video_rule.set_attribute_filter_function(
'data', lambda text: 'players.push' in text)
parser.add_rule(video_rule)
#
gallery_href_rule = ParserRule()
gallery_href_rule.add_activate_rule_level([('ul', 'class',
'video-tag-list')])
gallery_href_rule.add_process_rule_level('a', {'href'})
gallery_href_rule.set_attribute_modifier_function(
'href', lambda x: self.get_href(x, base_url))
parser.add_rule(gallery_href_rule)
self.proceed_parcing(parser, fname)
result = ParseResult()
if video_rule.is_result(): # len(video_rule.get_result()) > 0:
script = video_rule.get_result()[0]['data'].replace(' ',
'').replace(
'\\', '')
sources = script.partition("'sources':{")[2].partition(
'}')[0].split(',')
urls = list()
for item in sources:
part = item.strip("\n\t'").partition("':'")
if part[2].startswith('http://'):
data = dict(text=part[0],
url=URL(part[2].strip("'") + '*'))
urls.append(data)
if len(urls) == 1:
video = MediaData(urls[0]['url'])
elif len(urls) > 1:
video = MediaData(urls[-1]['url'])
for item in urls:
video.add_alternate(item)
else:
return result
result.set_type('video')
result.set_video(video)
for f in gallery_href_rule.get_result(['data', 'href']):
result.add_control(
ControlInfo(f['data'].strip(), URL(f['href'])))
return result
def add_key(old, key, value):
(addr, br, keys) = old.partition('?')
print(addr, br, keys)
pairs = keys.split('&')
print(pairs)
keys = ''
found = False
for pair in pairs:
if pair.startswith(key):
keys += key + '=' + value + '&'
found = True
else:
keys += pair + '&'
if not found:
keys += key + '=' + value
return addr + '?' + keys.strip('&')
def add_pages_info_to_result(rule, description_key='data-query-value'):
for item in rule.get_result(['data-query-key',
'data-query-value']):
print(item)
key = item['data-query-key']
val = item['data-query-value']
description = item[description_key].strip('\t')
old = base_url.get()
addr = add_key(old, key, val)
result.add_page(ControlInfo(description, URL(addr + '*')))
if channels_rule.is_result():
result.set_type('hrefs')
for item in channels_rule.get_result():
# print(item)
info = item['href'].rpartition('/')[2].strip('*')
result.add_thumb(
ThumbInfo(thumb_url=URL(item['src']),
href=URL(item['href']),
popup=info))
add_pages_info_to_result(channel_categories_rule,
description_key='data')
# for item in channel_categories_rule.get_result(['data-query-key', 'data-query-value']):
# print(item)
return result
if startpage_rule.is_result(): # len(startpage_rule.get_result()) > 0:
result.set_type('hrefs')
for item in startpage_rule.get_result(['href']):
# print(item)
t_url = item.get('data-original', item['src'])
result.add_thumb(
ThumbInfo(thumb_url=URL(t_url),
href=URL(item['href']),
popup=item.get('alt', '')))
add_pages_info_to_result(startpage_pages_rule)
return result
def parse_index_file(self, fname, base_url=URL()):
parser = SiteParser()
def star_get_url(txt=''):
return txt.partition('(')[2].partition(')')[0]
startpage_rule = ParserRule()
startpage_rule.add_activate_rule_level([('div', 'class', 'videos_form')])
startpage_rule.add_process_rule_level('a', {'href'})
startpage_rule.add_process_rule_level('img', {'data-lazy-src'})
startpage_rule.set_attribute_modifier_function('href', lambda x: self.get_href(x, base_url))
parser.add_rule(startpage_rule)
startpage_pages_rule = ParserRule()
startpage_pages_rule.add_activate_rule_level([('div', 'class', 'wp-pagenavi')])
# startpage_pages_rule.add_activate_rule_level([('a', 'class', 'current')])
startpage_pages_rule.add_process_rule_level('a', {'href'})
startpage_pages_rule.set_attribute_modifier_function('href', lambda x: self.get_href(x, base_url))
parser.add_rule(startpage_pages_rule)
startpage_hrefs_rule = ParserRule()
startpage_hrefs_rule.add_activate_rule_level([('ul', 'class', 'list')])
# startpage_hrefs_rule.add_activate_rule_level([('a', 'class', 'current')])
startpage_hrefs_rule.add_process_rule_level('a', {'href'})
startpage_hrefs_rule.set_attribute_modifier_function('href', lambda x: self.get_href(x, base_url))
parser.add_rule(startpage_hrefs_rule)
#
video_rule = ParserRule()
video_rule.add_activate_rule_level([('div', 'id', 'videos_page')])
video_rule.add_process_rule_level('script', {})
video_rule.set_attribute_filter_function('data', lambda text: 'jwplayer(' in text)
parser.add_rule(video_rule)
#
gallery_href_rule = ParserRule()
gallery_href_rule.add_activate_rule_level([('div', 'id', 'Categories')])
gallery_href_rule.add_process_rule_level('a', {'href'})
gallery_href_rule.set_attribute_modifier_function('href', lambda x: self.get_href(x, base_url))
parser.add_rule(gallery_href_rule)
self.proceed_parcing(parser, fname)
result = ParseResult()
if video_rule.is_result():
urls = UrlList()
for item in video_rule.get_result():
file = self.quotes(item['data'].replace(' ', ''), "file:'", "'")
urls.add('default', URL(file + '*'))
result.set_video(urls.get_media_data())
for f in gallery_href_rule.get_result(['data', 'href']):
result.add_control(ControlInfo(f['data'], URL(f['href'])))
return result
if startpage_rule.is_result(): # len(startpage_rule.get_result()) > 0:
result.set_type('hrefs')
for item in startpage_rule.get_result(['href']):
# print(item)
result.add_thumb(ThumbInfo(thumb_url=URL(item['data-lazy-src']), href=URL(item['href']),
popup=item.get('alt', '')))
for item in startpage_pages_rule.get_result(['href', 'data']):
result.add_page(ControlInfo(item['data'], URL(item['href'])))
for item in startpage_hrefs_rule.get_result(['href']):
href = item['href']
label = href.split('/')[-2]
# print(label,href)
result.add_control(ControlInfo(label, URL(href)))
return result
def parse_index_file(self, fname, base_url=URL()):
print('This site is very slow, be patient')
parser = SiteParser()
def star_get_url(txt=''):
return txt.partition('(')[2].partition(')')[0]
startpage_rule = ParserRule()
startpage_rule.add_activate_rule_level([('div', 'class', 'item ')])
startpage_rule.add_process_rule_level('a', {'href', 'title'})
startpage_rule.add_process_rule_level('img', {'src', 'alt'})
startpage_rule.set_attribute_modifier_function('href', lambda x: self.get_href(x, base_url))
parser.add_rule(startpage_rule)
startpage_pages_rule = ParserRule()
startpage_pages_rule.add_activate_rule_level([('div', 'class', 'pagination-holder')])
# startpage_pages_rule.add_activate_rule_level([('a', 'class', 'current')])
startpage_pages_rule.add_process_rule_level('a', {'href'})
startpage_pages_rule.set_attribute_modifier_function('href', lambda x: self.get_href(x, base_url))
parser.add_rule(startpage_pages_rule)
startpage_hrefs_rule = ParserRule()
startpage_hrefs_rule.add_activate_rule_level([('ul', 'class', 'list')])
# startpage_hrefs_rule.add_activate_rule_level([('a', 'class', 'current')])
startpage_hrefs_rule.add_process_rule_level('a', {'href'})
startpage_hrefs_rule.set_attribute_modifier_function('href', lambda x: self.get_href(x, base_url))
parser.add_rule(startpage_hrefs_rule)
#
video_rule = ParserRule()
video_rule.add_activate_rule_level([('div', 'class', 'player')])
video_rule.add_process_rule_level('script', {})
video_rule.set_attribute_filter_function('data', lambda text: 'flashvars' in text)
parser.add_rule(video_rule)
#
gallery_href_rule = ParserRule()
gallery_href_rule.add_activate_rule_level([('div', 'class', 'item')])
gallery_href_rule.add_process_rule_level('a', {'href'})
gallery_href_rule.set_attribute_modifier_function('href', lambda x: self.get_href(x, base_url))
parser.add_rule(gallery_href_rule)
self.proceed_parcing(parser, fname)
result = ParseResult()
if video_rule.is_result():
urls = UrlList()
for item in video_rule.get_result():
file = self.quotes(item['data'].replace(' ', ''), "video_url:'", "'")
urls.add('default', URL(file))
result.set_video(urls.get_media_data())
for f in gallery_href_rule.get_result(['data', 'href']):
result.add_control(ControlInfo(f['data'], URL(f['href'])))
return result
if startpage_rule.is_result():
for item in startpage_rule.get_result(['href']):
result.add_thumb(
ThumbInfo(thumb_url=URL(item['src']), href=URL(item['href']), popup=item.get('alt', '')))
for item in startpage_pages_rule.get_result(['href', 'data']):
result.add_page(ControlInfo(item['data'], URL(item['href'])))
for item in startpage_hrefs_rule.get_result(['href']):
href = item['href']
label = href.split('/')[-2]
# print(label,href)
result.add_control(ControlInfo(label, URL(href)))
return result
def parse_index_file(self, fname, base_url=URL()):
parser = SiteParser()
startpage_rule = ParserRule()
# startpage_rule.add_activate_rule_level([('section', '', '')])
startpage_rule.add_activate_rule_level([('div', 'class', 'inner-block')
])
startpage_rule.add_process_rule_level('a', {'href'})
startpage_rule.add_process_rule_level('img', {'src', 'alt'})
startpage_rule.set_attribute_modifier_function(
'href', lambda x: self.get_href(x, base_url))
startpage_rule.set_attribute_modifier_function(
'src', lambda x: self.get_href(x, base_url))
parser.add_rule(startpage_rule)
startpage_pages_rule = ParserRule()
startpage_pages_rule.add_activate_rule_level([('div', 'id',
'pagination')])
startpage_pages_rule.add_process_rule_level('a', {'href'})
startpage_pages_rule.set_attribute_modifier_function(
'href', lambda x: self.get_href(x, base_url))
parser.add_rule(startpage_pages_rule)
video_rule = ParserRule()
video_rule.add_activate_rule_level([('div', 'class', 'stage-video')])
video_rule.add_process_rule_level('script', {})
video_rule.set_attribute_filter_function(
'data', lambda text: 'jwplayer(' in text)
parser.add_rule(video_rule)
#
gallery_href_rule = ParserRule()
gallery_href_rule.add_activate_rule_level([
('ul', 'class', 'stats-list stats-list--plain')
])
gallery_href_rule.add_process_rule_level('a', {'href'})
gallery_href_rule.set_attribute_modifier_function(
'href', lambda x: self.get_href(x, base_url))
parser.add_rule(gallery_href_rule)
self.proceed_parcing(parser, fname)
result = ParseResult()
if video_rule.is_result():
urls = UrlList()
for item in video_rule.get_result():
# print(item['data'])
file = self.quotes(item['data'].replace(' ', ''), 'file:"',
'"')
urls.add('DEFAULT', URL(file))
result.set_video(urls.get_media_data())
for f in gallery_href_rule.get_result(['data', 'href']):
if '/user/' in f['href']:
form = '"{0}"'
else:
form = '{0}'
result.add_control(
ControlInfo(form.format(f['data']), URL(f['href'])))
return result
if startpage_rule.is_result():
for item in startpage_rule.get_result(['href']):
result.add_thumb(
ThumbInfo(thumb_url=URL(item['src']),
href=URL(item['href']),
popup=item.get('alt'
'')))
for item in startpage_pages_rule.get_result(['href', 'data']):
result.add_page(ControlInfo(item['data'], URL(item['href'])))
return result
def parse_index_file(self, fname, base_url=URL()):
parser = SiteParser()
startpage_rule = ParserRule()
startpage_rule.add_activate_rule_level([('div', 'class', 'item '),
('div', 'class',
'list-categories')])
startpage_rule.add_process_rule_level('a', {'href', 'title'})
startpage_rule.add_process_rule_level('img', {'data-original', 'alt'})
startpage_rule.set_attribute_modifier_function(
'href', lambda x: self.get_href(x, base_url))
startpage_rule.set_attribute_modifier_function(
'src', lambda x: self.get_href(x, base_url))
parser.add_rule(startpage_rule)
startpage_pages_rule = ParserRule()
startpage_pages_rule.add_activate_rule_level([('div', 'class',
'pagination')])
# startpage_pages_rule.add_activate_rule_level([('a', 'class', 'current')])
startpage_pages_rule.add_process_rule_level('li', {'class'})
startpage_pages_rule.add_process_rule_level('a',
{'href', 'data-query'})
startpage_pages_rule.set_attribute_filter_function(
'class', lambda x: x == 'page')
startpage_pages_rule.set_attribute_modifier_function(
'data-query', lambda x: x.partition(':')[2])
startpage_pages_rule.set_attribute_modifier_function(
'href', lambda x: self.get_href(x, base_url))
parser.add_rule(startpage_pages_rule)
#
video_rule = ParserRule()
video_rule.add_activate_rule_level([('div', 'class', 'player')])
video_rule.add_process_rule_level('script', {})
video_rule.set_attribute_filter_function(
'data', lambda text: 'flashvars' in text)
# video_rule.set_attribute_modifier_function('src',lambda x:self.get_href(x,base_url))
parser.add_rule(video_rule)
#
gallery_href_rule = ParserRule()
gallery_href_rule.add_activate_rule_level([('div', 'class', 'info')])
gallery_href_rule.add_process_rule_level('a', {'href'})
# gallery_href_rule.set_attribute_filter_function('href',lambda x:'/profiles/' not in x)
gallery_href_rule.set_attribute_modifier_function(
'href', lambda x: self.get_href(x, base_url))
parser.add_rule(gallery_href_rule)
gallery_user_rule = ParserRule()
gallery_user_rule.add_activate_rule_level([('div', 'class',
'block-user')])
gallery_user_rule.add_process_rule_level('a', {'href', 'title'})
# gallery_user_rule.set_attribute_filter_function('href',lambda x:'/profiles/' in x)
gallery_user_rule.set_attribute_modifier_function(
'href', lambda x: self.get_href(x + 'videos/', base_url))
parser.add_rule(gallery_user_rule)
self.proceed_parcing(parser, fname)
result = ParseResult()
if video_rule.is_result():
urls = UrlList()
for item in video_rule.get_result():
flashvars = self.quotes(
item['data'].replace(' ',
'').replace('\n',
'').replace('\t', ''),
'flashvars={', '};').split(',')
fv = dict()
for flashvar in flashvars:
split = flashvar.partition(':')
fv[split[0]] = split[2].strip("'\"")
# file=self.quotes(item['data'],'file:',',').strip(' "')
urls.add('default', URL(fv['video_url'] + '*'))
result.set_video(urls.get_media_data())
for f in gallery_user_rule.get_result(['title', 'href']):
result.add_control(
ControlInfo('"' + f['title'].strip() + '"',
URL(f['href'])))
for f in gallery_href_rule.get_result(['data', 'href']):
result.add_control(ControlInfo(f['data'], URL(f['href'])))
return result
if startpage_rule.is_result():
for item in startpage_rule.get_result(['href']):
result.add_thumb(
ThumbInfo(thumb_url=URL(item['data-original']),
href=URL(item['href']),
popup=item.get('alt', item.get('title', ''))))
for item in startpage_pages_rule.get_result(['href',
'data-query']):
result.add_page(
ControlInfo(item['data-query'], URL(item['href'])))
return result
def parse_index_file(self, fname, base_url=URL()):
parser = SiteParser()
startpage_rule = ParserRule()
startpage_rule.add_activate_rule_level([('div', 'class',
'image-delete'),
('div', 'class', 'thumbs'),
('div', 'class', 'image')])
startpage_rule.add_process_rule_level('a', {'href'})
startpage_rule.add_process_rule_level('img', {'src', 'alt'})
parser.add_rule(startpage_rule)
startpage_pages_rule = ParserRule()
startpage_pages_rule.add_activate_rule_level([('div', 'class', 'pager')
])
startpage_pages_rule.add_process_rule_level('a', {'href'})
startpage_pages_rule.set_attribute_modifier_function(
'href', lambda x: base_url.domain() + x)
parser.add_rule(startpage_pages_rule)
video_rule = ParserRule()
video_rule.add_activate_rule_level([('div', 'class', 'main')])
video_rule.add_process_rule_level('script', {})
video_rule.set_attribute_filter_function(
'data', lambda text: 'function getEmbed()' in text)
parser.add_rule(video_rule)
picture_rule = ParserRule()
picture_rule.add_activate_rule_level([('div', 'class', 'thumbs2')])
picture_rule.add_process_rule_level('a', {'href'})
picture_rule.add_process_rule_level('img', {'src'})
# picture_rule.set_attribute_modifier_function('src', lambda text: text.replace('thumb', 'origin'))
parser.add_rule(picture_rule)
picture_tags_rule = ParserRule()
picture_tags_rule.add_activate_rule_level([('div', 'class', 'main')])
picture_tags_rule.add_process_rule_level('a', {'href'})
picture_tags_rule.set_attribute_filter_function(
'href', lambda txt: '/categories/' in txt or '/model/' in txt)
parser.add_rule(picture_tags_rule)
for s in open(fname, encoding='utf-8'):
parser.feed(s)
result = ParseResult()
if len(video_rule.get_result()) > 0:
result.set_video(
MediaData(
URL(
self.get_attr_from_script(
video_rule.get_result()[0]['data']))))
result.set_type('video')
for item in picture_tags_rule.get_result(['href', 'data']):
result.add_control(ControlInfo(item['data'],
URL(item['href'])))
return result
if len(startpage_rule.get_result()) > 0:
result.set_type('hrefs')
for item in startpage_rule.get_result():
result.add_thumb(
ThumbInfo(thumb_url=URL(item['src']),
href=URL(item['href']),
popup=item.get('alt', '')))
for item in startpage_pages_rule.get_result(['href', 'data']):
result.add_page(ControlInfo(item['data'], URL(item['href'])))
if len(picture_rule.get_result()) > 0:
result.set_type('pictures')
result.set_picture_collector(ELSitePictureCollector())
i = 1
for f in picture_rule.get_result():
result.add_full(
FullPictureInfo(abs_href=URL(f['href']),
rel_name='%03d.jpg' % i))
i += 1
for item in picture_tags_rule.get_result(['href', 'data']):
result.add_control(ControlInfo(item['data'],
URL(item['href'])))
return result
def parse_index_file(self, fname, base_url=URL()):
parser = SiteParser()
domain = base_url.domain()
href_rule = ParserRule() # startpage & model's page
href_rule.add_activate_rule_level([('div', 'id', 'lst-galleries'),
('div', 'class', 'lblock'),
('div', 'class', 'modal_info_full')
])
href_rule.add_process_rule_level('a', {'href'})
href_rule.add_process_rule_level('img', {'src', 'alt'})
href_rule.set_attribute_modifier_function(
'href', lambda x: self.get_href(x, base_url))
href_rule.set_attribute_modifier_function(
'src', lambda x: self.get_href(x, base_url))
href_rule.set_attribute_filter_function('href', self.thumb_href_filter)
href_rule.set_attribute_filter_function('src', self.thumb_src_filter)
parser.add_rule(href_rule)
href_page_rule = ParserRule() # page number in model's page
href_page_rule.add_activate_rule_level([('div', 'class', 'pages'),
('div', 'class', 'cat')])
href_page_rule.add_process_rule_level('a', {'href'})
href_page_rule.set_attribute_modifier_function(
'href', lambda x: self.get_href(x, base_url))
parser.add_rule(href_page_rule)
model_litera_rule = ParserRule()
model_litera_rule.add_activate_rule_level([('div', 'class',
'babe_index')])
model_litera_rule.add_process_rule_level('a', {'href'})
model_litera_rule.set_attribute_modifier_function(
'href', lambda x: self.get_href(x, base_url))
parser.add_rule(model_litera_rule)
model_more_rule = ParserRule()
model_more_rule.add_activate_rule_level([('div', 'class', 'more'),
('div', 'id', 'MoreCont')])
model_more_rule.add_process_rule_level('a', {'href'})
model_more_rule.set_attribute_modifier_function(
'href', lambda x: self.get_href(x, base_url))
model_more_rule.set_attribute_filter_function('href',
self.thumb_href_filter)
parser.add_rule(model_more_rule)
picture_rule = ParserRule() # gallery rule
picture_rule.add_activate_rule_level([('div', 'class', 'lblock')])
# picture_rule.add_activate_rule_level([('ul', 'class', 'block')])
picture_rule.add_process_rule_level('a', {'href'})
picture_rule.add_process_rule_level('img', {'alt'})
picture_rule.set_attribute_modifier_function(
'href', lambda x: self.get_href(x, base_url))
picture_rule.set_attribute_filter_function(
'href', lambda x: x.endswith('.jpg'))
parser.add_rule(picture_rule)
picture_href_rule = ParserRule() # gallery href's rule
picture_href_rule.add_activate_rule_level([('div', 'id', 'ModelMenu'),
('div', 'class', 'lblock')])
picture_href_rule.add_process_rule_level('a', {'href', 'title'})
picture_href_rule.set_attribute_modifier_function(
'href', lambda x: self.get_href(x, base_url))
picture_href_rule.set_attribute_filter_function(
'href', self.gal_href_filter)
parser.add_rule(picture_href_rule)
for s in open(fname):
parser.feed(s)
result = ParseResult()
if len(picture_rule.get_result()) > 0:
result.set_type('pictures')
i = 1
for f in picture_rule.get_result():
x = FullPictureInfo(abs_href=URL(f['href']),
rel_name='%03d.jpg' % i)
result.add_full(x)
i += 1
for f in picture_href_rule.get_result(['href', 'title']):
# print(f)
result.add_control(
ControlInfo(text=f['title'], url=URL(f['href'])))
return result
if len(href_rule.get_result()) > 0:
result.set_type('hrefs')
for item in href_rule.get_result():
result.add_thumb(
ThumbInfo(thumb_url=URL(item['src']),
href=URL(item['href']),
popup=item.get('alt', '')))
for item in model_more_rule.get_result(['href', 'data']):
result.add_control(ControlInfo(item['data'],
URL(item['href'])))
for item in model_litera_rule.get_result(['href', 'data']):
result.add_control(ControlInfo(item['data'],
URL(item['href'])))
for item in href_page_rule.get_result(['href', 'data']):
result.add_page(ControlInfo(item['data'], URL(item['href'])))
return result
def parse_index_file(self, fname, base_url=URL()):
parser = SiteParser()
def star_get_url(txt=''):
return txt.partition('(')[2].partition(')')[0]
startpage_rule = ParserRule()
startpage_rule.add_activate_rule_level([('div', 'class', 'thumb')])
startpage_rule.add_process_rule_level('a', {'href', 'title'})
startpage_rule.add_process_rule_level('img', {'src', 'data-original'})
startpage_rule.set_attribute_modifier_function(
'href', lambda x: base_url.domain() + x)
parser.add_rule(startpage_rule)
startpage_pages_rule = ParserRule()
startpage_pages_rule.add_activate_rule_level([('ul', 'class',
'pagination')])
# startpage_pages_rule.add_activate_rule_level([('a', 'class', 'current')])
startpage_pages_rule.add_process_rule_level('a', {'href'})
startpage_pages_rule.set_attribute_modifier_function(
'href', lambda x: base_url.domain() + x + '*')
parser.add_rule(startpage_pages_rule)
startpage_hrefs_rule = ParserRule()
startpage_hrefs_rule.add_activate_rule_level([('div', 'class',
'listTags listTags5')])
# startpage_hrefs_rule.add_activate_rule_level([('a', 'class', 'current')])
startpage_hrefs_rule.add_process_rule_level('a', {'href'})
startpage_hrefs_rule.set_attribute_modifier_function(
'href', lambda x: base_url.domain() + x + '*')
parser.add_rule(startpage_hrefs_rule)
#
video_rule = ParserRule()
video_rule.add_activate_rule_level([('body', '', '')])
video_rule.add_process_rule_level('script', {})
video_rule.set_attribute_filter_function(
'data', lambda text: 'jwplayer("jw_video").setup' in text)
parser.add_rule(video_rule)
#
gallery_href_rule = ParserRule()
gallery_href_rule.add_activate_rule_level([('ul', 'class',
'stickerNav')])
gallery_href_rule.add_process_rule_level('a', {'href'})
gallery_href_rule.set_attribute_modifier_function(
'href', lambda x: base_url.domain() + x + '*')
parser.add_rule(gallery_href_rule)
#
# gallery_channel_rule = ParserRule()
# gallery_channel_rule.add_activate_rule_level([('p', 'class', 'source')])
# gallery_channel_rule.add_process_rule_level('a', {'href'})
# gallery_channel_rule.set_attribute_modifier_function('href', lambda x: base_url.domain() + x + '*')
# parser.add_rule(gallery_channel_rule)
self.proceed_parcing(parser, fname)
result = ParseResult()
if len(video_rule.get_result()) > 0:
script = video_rule.get_result()[0][
'data'] # .replace(' ', '').replace('\\','')
# print(script)
# print('len=',len(video_rule.get_result()))
sources = script.partition('"sources":[{')[2].partition(
'}]')[0].split('},{')
# for i in sources:
# print(i)
def parce(txt):
label = txt.partition('"label":"')[2].partition('"')[0]
file = txt.partition('"file":"')[2].partition('"')[0]
# print(label,file)
return dict(text=label, url=URL(file + '*'))
if len(sources) == 1:
video = MediaData(parce(sources[0])['url'])
elif len(sources) > 1:
video = MediaData(parce(sources[len(sources) - 1])['url'])
for item in sources:
video.add_alternate(parce(item))
else:
return result
result.set_type('video')
result.set_video(video)
#
# for f in gallery_channel_rule.get_result(['data', 'href']):
# result.add_control(ControlInfo(f['data'], URL(f['href'])))
for f in gallery_href_rule.get_result(['data', 'href']):
result.add_control(ControlInfo(f['data'], URL(f['href'])))
return result
if len(startpage_rule.get_result()) > 0:
result.set_type('hrefs')
for item in startpage_rule.get_result(['href', 'data-original']):
result.add_thumb(
ThumbInfo(thumb_url=URL(item['data-original']),
href=URL(item['href']),
popup=item.get('title', '')))
for item in startpage_pages_rule.get_result(['href', 'data']):
result.add_page(ControlInfo(item['data'], URL(item['href'])))
if len(startpage_hrefs_rule.get_result(['href', 'data'])) > 0:
for item in startpage_hrefs_rule.get_result(['href', 'data']):
result.add_control(
ControlInfo(item['data'], URL(item['href'])))
return result
def parse_index_file(self, fname, base_url=URL()):
parser = SiteParser()
def star_get_url(txt=''):
return txt.partition('(')[2].partition(')')[0]
startpage_rule = ParserRule()
startpage_rule.add_activate_rule_level([('div', 'class',
'thmb-wrapper')])
startpage_rule.add_process_rule_level('a', {'href'})
startpage_rule.add_process_rule_level(
'img', {'src', 'alt', 'data-src', 'data-original'})
startpage_rule.set_attribute_modifier_function(
'href', lambda x: base_url.domain() + x)
parser.add_rule(startpage_rule)
startpage_pages_rule = ParserRule()
startpage_pages_rule.add_activate_rule_level([('div', 'id',
'divPagination')])
# startpage_pages_rule.add_activate_rule_level([('a', 'class', 'current')])
startpage_pages_rule.add_process_rule_level('a', {'href'})
startpage_pages_rule.set_attribute_modifier_function(
'href', lambda x: base_url.domain() + x + '*')
parser.add_rule(startpage_pages_rule)
startpage_hrefs_rule = ParserRule()
startpage_hrefs_rule.add_activate_rule_level([('div', 'class',
'listTags listTags5')])
# startpage_hrefs_rule.add_activate_rule_level([('a', 'class', 'current')])
startpage_hrefs_rule.add_process_rule_level('a', {'href'})
startpage_hrefs_rule.set_attribute_modifier_function(
'href', lambda x: base_url.domain() + x + '*')
parser.add_rule(startpage_hrefs_rule)
#
video_rule = ParserRule()
video_rule.add_activate_rule_level([('div', 'id', 'videoContainer')])
video_rule.add_process_rule_level('script', {})
video_rule.set_attribute_filter_function(
'data', lambda text: 'playerData.cdnPath' in text)
parser.add_rule(video_rule)
#
gallery_href_rule = ParserRule()
gallery_href_rule.add_activate_rule_level([
('div', 'class', 'video-info-tags float-left')
])
gallery_href_rule.add_process_rule_level('a', {'href'})
gallery_href_rule.set_attribute_modifier_function(
'href', lambda x: base_url.domain() + x + '*')
parser.add_rule(gallery_href_rule)
gallery_channel_rule = ParserRule()
gallery_channel_rule.add_activate_rule_level([
('div', 'class', 'video-info-uploaded float-right')
])
gallery_channel_rule.add_process_rule_level('a', {'href'})
gallery_channel_rule.set_attribute_modifier_function(
'href', lambda x: base_url.domain() + x + '*')
gallery_channel_rule.set_attribute_filter_function(
'href', lambda x: '/categories/' in x)
parser.add_rule(gallery_channel_rule)
gallery_user_rule = ParserRule()
gallery_user_rule.add_activate_rule_level([
('div', 'class', 'video-info-uploaded float-right')
])
gallery_user_rule.add_process_rule_level('a', {'href'})
gallery_user_rule.set_attribute_modifier_function(
'href', lambda x: base_url.domain() + x + '*')
gallery_user_rule.set_attribute_filter_function(
'href', lambda x: '/user/' in x)
parser.add_rule(gallery_user_rule)
self.proceed_parcing(parser, fname)
result = ParseResult()
if video_rule.is_result(): # len(video_rule.get_result()) > 0:
script = video_rule.get_result()[0]['data'].replace(
' ', '') # .replace('\\','')
# print(video_rule.get_result()[0]['data'])
# print(script)
# print('len=',len(video_rule.get_result()))
lines = script.split('\n')
data = list()
for i in lines:
if i.strip().startswith('playerData.cdnPath'):
if "''" not in i:
data.append(i.strip())
# print(i.strip())
def parce(txt):
label = txt.partition('playerData.cdnPath')[2].partition(
'=')[0]
file = txt.partition("'")[2].partition("'")[0].replace(
'%3A', ':').replace('%2F', '/').replace('%26', '&')
# print(label,file)
return dict(text=label, url=URL(file + '*'))
if len(data) == 1:
video = MediaData(parce(data[0])['url'])
elif len(data) > 1:
video = MediaData(parce(data[len(data) - 1])['url'])
for item in data:
video.add_alternate(parce(item))
else:
return result
result.set_type('video')
result.set_video(video)
for f in gallery_user_rule.get_result(['data', 'href']):
# print(f)
result.add_control(
ControlInfo('"' + f['data'] + '"', URL(f['href'])))
for f in gallery_channel_rule.get_result(['data', 'href']):
result.add_control(ControlInfo(f['data'], URL(f['href'])))
for f in gallery_href_rule.get_result(['data', 'href']):
result.add_control(ControlInfo(f['data'], URL(f['href'])))
return result
if startpage_rule.is_result(): # len(startpage_rule.get_result()) > 0:
result.set_type('hrefs')
for item in startpage_rule.get_result(['href']):
# print(item)
if 'data-src' in item.keys():
src = item['data-src']
elif 'data-original' in item.keys():
src = item['data-original']
elif 'src' in item.keys():
src = item['src']
else:
print('New key found. Need rewrite "startpage_rule"')
continue
# print(src,item.get('src',''),item.get('data-src',''))
result.add_thumb(
ThumbInfo(thumb_url=URL(src),
href=URL(item['href']),
popup=item.get('alt', '')))
for item in startpage_pages_rule.get_result(['href', 'data']):
result.add_page(ControlInfo(item['data'], URL(item['href'])))
if len(startpage_hrefs_rule.get_result(['href', 'data'])) > 0:
for item in startpage_hrefs_rule.get_result(['href', 'data']):
result.add_control(
ControlInfo(item['data'], URL(item['href'])))
return result
def parse_index_file(self, fname, base_url=URL()):
parser = SiteParser()
startpage_rule = ParserRule()
startpage_rule.add_activate_rule_level([('div', 'class', 'item ')])
startpage_rule.add_process_rule_level('a', {'href', 'title'})
startpage_rule.add_process_rule_level('img', {'data-original', 'alt'})
startpage_rule.set_attribute_modifier_function(
'href', lambda x: self.get_href(x, base_url))
startpage_rule.set_attribute_modifier_function(
'src', lambda x: self.get_href(x, base_url))
parser.add_rule(startpage_rule)
startpage_pages_rule = ParserRule()
startpage_pages_rule.add_activate_rule_level([('div', 'class',
'pagination-holder')])
# startpage_pages_rule.add_activate_rule_level([('a', 'class', 'current')])
startpage_pages_rule.add_process_rule_level('a', {'href'})
startpage_pages_rule.set_attribute_modifier_function(
'href', lambda x: self.get_href(x, base_url))
parser.add_rule(startpage_pages_rule)
video_rule = ParserRule()
video_rule.add_activate_rule_level([('div', 'class', 'player')])
video_rule.add_process_rule_level('script', {})
video_rule.set_attribute_filter_function(
'data', lambda text: 'flashvars' in text)
# video_rule.set_attribute_modifier_function('src',lambda x:self.get_href(x,base_url))
parser.add_rule(video_rule)
#
gallery_href_rule = ParserRule()
gallery_href_rule.add_activate_rule_level([('div', 'class', 'info')])
gallery_href_rule.add_process_rule_level('a', {'href'})
gallery_href_rule.set_attribute_modifier_function(
'href', lambda x: self.get_href(x, base_url))
parser.add_rule(gallery_href_rule)
self.proceed_parcing(parser, fname)
result = ParseResult()
if video_rule.is_result():
urls = UrlList()
for item in video_rule.get_result():
flashvars = self.quotes(
item['data'].replace(' ',
'').replace('\n',
'').replace('\t', ''),
'flashvars={', '};').split(',')
fv = dict()
for flashvar in flashvars:
split = flashvar.partition(':')
fv[split[0]] = split[2].strip("'\"")
files = dict()
for f in fv:
if fv[f].startswith('http://') and fv[f].rpartition(
'.')[2].strip('/') == 'mp4':
file = fv[f]
label = fv.get(f + '_text', f)
files[label] = file
for key in sorted(files.keys(), reverse=True):
urls.add(key, URL(files[key]))
result.set_video(urls.get_media_data())
for f in gallery_href_rule.get_result(['data', 'href']):
result.add_control(ControlInfo(f['data'], URL(f['href'])))
return result
if startpage_rule.is_result():
for item in startpage_rule.get_result(['href']):
result.add_thumb(
ThumbInfo(thumb_url=URL(item['data-original']),
href=URL(item['href']),
popup=item.get('alt', item.get('title', ''))))
for item in startpage_pages_rule.get_result(['href', 'data']):
result.add_page(ControlInfo(item['data'], URL(item['href'])))
return result
def parse_index_file(self, fname, base_url=URL()):
if base_url.contain('format=json'):
xhr_page = True
else:
xhr_page = False
parser = SiteParser()
startpage_rule = ParserRule()
startpage_rule.add_activate_rule_level([
('div', 'class',
'relative margin-auto video-box-wrapper normal-box')
])
startpage_rule.add_process_rule_level('a', {'href'})
startpage_rule.add_process_rule_level('img', {'alt', 'data-srcmedium'})
startpage_rule.set_attribute_modifier_function(
'href', lambda x: self.get_href(x, base_url))
parser.add_rule(startpage_rule)
video_rule = ParserRule()
video_rule.add_activate_rule_level([('div', 'class', 'video-container')
])
video_rule.add_process_rule_level('script', {})
video_rule.set_attribute_filter_function(
'data', lambda text: 'flashvars=' in text.replace(' ', ''))
# video_rule.set_attribute_modifier_function('src', lambda x: self.get_href(x, base_url))
parser.add_rule(video_rule)
#
video_href_rule = ParserRule()
video_href_rule.add_activate_rule_level([('div', 'class',
'ibInfo js_ibInfo')])
video_href_rule.add_process_rule_level('a', {'href'})
video_href_rule.set_attribute_filter_function(
'href', lambda x: 'javascript' not in x)
video_href_rule.set_attribute_modifier_function(
'href', lambda x: self.get_href(x, base_url))
parser.add_rule(video_href_rule)
video_user_rule = ParserRule()
video_user_rule.add_activate_rule_level([('div', 'class', 'ibLine1')])
video_user_rule.add_process_rule_level('a', {'href'})
# video_user_rule.set_attribute_filter_function('href',lambda x: 'javascript' not in x)
video_user_rule.set_attribute_modifier_function(
'href', lambda x: self.get_href(x, base_url))
parser.add_rule(video_user_rule)
if not xhr_page:
self.proceed_parcing(parser, fname)
result = ParseResult()
if video_rule.is_result():
urls = UrlList()
for item in video_rule.get_result():
flashvars = self.quotes(item['data'].replace(' ', ''),
'flashvars={',
'};').replace('\\', '').split(',"')
for v in flashvars:
if v.startswith('quality_'):
label = self.quotes(v, 'quality_', '"')
file = self.quotes(v, ':"', '"')
urls.add(label, URL(file + '*'))
result.set_video(urls.get_media_data(-1))
for f in video_user_rule.get_result(['data', 'href']):
result.add_control(
ControlInfo('"' + f['data'] + '"', URL(f['href'])))
for f in video_href_rule.get_result(['data', 'href']):
result.add_control(ControlInfo(f['data'], URL(f['href'])))
return result
if xhr_page:
def parce_data(data, buttons_added, base_url, items_name='items'):
xhr_data = base_url.xhr_data
items = data[items_name]
nav = data['navigation']
last_page = nav['lastPage']
if last_page is None: last_page = 1
curr_page = int(base_url.get().rpartition('page=')[2])
pattern = nav['urlPattern'].replace('[%pageId%]', '{}') + '*'
for item in items:
thumb_url = item['thumb_url']
title = item['specialchars_title']
url = item['video_link']
result.add_thumb(
ThumbInfo(thumb_url=URL(thumb_url),
href=URL(url + '*'),
popup=title))
# print('Page {} of {}'.format(curr_page, last_page), pattern.format(curr_page))
if not buttons_added:
result.add_page(ControlInfo('1', xhr_data['base_url']))
p_from = curr_page - 5
if p_from < 2: p_from = 2
p_to = curr_page + 5
if p_to > last_page: p_to = last_page
for x in range(p_from, p_to):
url = URL(pattern.format(x), xhr_data=xhr_data)
if x == curr_page: x = str(x) + '(this)'
result.add_page(ControlInfo(str(x), url))
last_url = URL(pattern.format(last_page),
xhr_data=xhr_data)
result.add_page(ControlInfo(str(last_page), last_url))
buttons_added = True
return buttons_added
with open(fname) as fp:
try:
json_data = json.load(fp)
buttons_added = False
if base_url.contain('/keyword/'):
for i in json_data:
buttons_added = parce_data(i, buttons_added,
base_url)
elif base_url.contain('/users/'):
parce_data(json_data['response'],
buttons_added,
base_url,
items_name='videos')
else:
for i in json_data:
buttons_added = parce_data(json_data[i],
buttons_added, base_url)
except ValueError:
pass
return result
if startpage_rule.is_result():
for item in startpage_rule.get_result(['href']):
result.add_thumb(
ThumbInfo(thumb_url=URL(item['data-srcmedium']),
href=URL(item['href']),
popup=item.get('alt', '')))
xhr_data = {'base_url': base_url}
next_url = URL(
base_url.get() + '*',
xhr_data=xhr_data) # +'?format=json&number_pages=1&page=2*'
next_url.add_query([('format', 'json'), ('number_pages', '1'),
('page', '2')])
result.add_page(ControlInfo('next', next_url))
return result
def parse_index_file(self, fname, base_url=URL()):
# print(base_url)
parser = SiteParser()
startpage_rule = ParserRule()
startpage_rule.add_activate_rule_level([('div', 'class', 'thumblinks')
])
# startpage_rule.add_process_rule_level('div', {})
startpage_rule.add_process_rule_level('a', {'href'})
startpage_rule.add_process_rule_level('img', {'src', 'alt'})
startpage_rule.set_attribute_modifier_function(
'href', lambda x: self.get_href(x, base_url) + '*')
startpage_rule.set_attribute_modifier_function(
'src', lambda x: self.get_href(x, base_url) + '*')
parser.add_rule(startpage_rule)
tags_rule = ParserRule()
tags_rule.add_activate_rule_level([('div', 'id', 'divTags')])
tags_rule.add_process_rule_level('a', {'href'})
tags_rule.set_attribute_modifier_function(
'href', lambda x: base_url.domain() + x)
parser.add_rule(tags_rule)
startpage_pages_rule = ParserRule()
startpage_pages_rule.add_activate_rule_level([('td', 'class', 'pages')
])
startpage_pages_rule.add_process_rule_level('a', {'href'})
startpage_pages_rule.set_attribute_modifier_function(
'href', lambda x: base_url.domain() + x)
parser.add_rule(startpage_pages_rule)
picture_base_addr_rule = ParserRule()
picture_base_addr_rule.add_activate_rule_level([('div', 'class',
'imagelinks')])
picture_base_addr_rule.add_process_rule_level('script', {})
picture_base_addr_rule.set_attribute_filter_function(
'data', lambda x: 'unescape' in x)
parser.add_rule(picture_base_addr_rule)
picture_rule = ParserRule()
picture_rule.add_activate_rule_level([('div', 'class', 'imagelinks')])
picture_rule.add_process_rule_level('script', {})
picture_rule.set_attribute_filter_function('data',
lambda x: "'src'" in x)
parser.add_rule(picture_rule)
for s in open(fname):
parser.feed(s)
result = ParseResult()
if len(picture_base_addr_rule.get_result()) > 0:
result.set_type('pictures')
base = \
picture_base_addr_rule.get_result()[0]['data'].replace('%2f', '/').partition("unescape('//")[2].partition(
"'")[0]
# print(base)
i = 1
for f in picture_rule.get_result():
picname = f['data'].partition("+'")[2].partition("'")[0]
# print(picname)
result.add_full(
FullPictureInfo(abs_href=URL(base + picname + '*'),
rel_name='%03d.jpg' % i))
i += 1
for item in tags_rule.get_result(['href', 'data']):
result.add_control(ControlInfo(item['data'],
URL(item['href'])))
return result
if len(startpage_rule.get_result()) > 0:
result.set_type('hrefs')
for item in startpage_rule.get_result():
# print(item)
result.add_thumb(
ThumbInfo(thumb_url=URL(item['src']),
href=URL(item['href']),
popup=item.get('alt', '')))
for item in tags_rule.get_result(['href', 'data']):
result.add_control(ControlInfo(item['data'],
URL(item['href'])))
for item in startpage_pages_rule.get_result(['href', 'data']):
result.add_page(ControlInfo(item['data'], URL(item['href'])))
return result
def parse_index_file(self, fname, base_url=URL()):
parser = SiteParser()
startpage_rule = ParserRule()
startpage_rule.add_activate_rule_level([('div', 'class',
'fixed-content')])
startpage_rule.add_process_rule_level('a', {'href', 'class'})
startpage_rule.add_process_rule_level('div', {'style'})
startpage_rule.set_attribute_filter_function(
'class', lambda x: x == 'thumbnail')
startpage_rule.set_attribute_modifier_function(
'href', lambda x: self.get_href(x, base_url))
startpage_rule.set_attribute_modifier_function(
'style', lambda x: x.partition("url('")[2].partition("')")[0])
parser.add_rule(startpage_rule)
startpage_pages_rule = ParserRule()
startpage_pages_rule.add_activate_rule_level([
('div', 'class', 'col-xs-12 content-pagination')
])
startpage_pages_rule.add_process_rule_level('a', {'href'})
startpage_pages_rule.set_attribute_modifier_function(
'href', lambda x: self.get_href(x, base_url))
parser.add_rule(startpage_pages_rule)
tags_rule = ParserRule()
tags_rule.add_activate_rule_level([('section', 'id', 'footer-tag')])
tags_rule.add_process_rule_level('a', {'href'})
tags_rule.set_attribute_modifier_function(
'href', lambda x: self.get_href(x, base_url))
parser.add_rule(tags_rule)
categories_rule = ParserRule()
categories_rule.add_activate_rule_level([('ul', 'class',
'nav navbar-nav')])
categories_rule.add_process_rule_level('a', {'href'})
categories_rule.set_attribute_filter_function(
'href', lambda x: '/Category/' in x and "#" not in x)
categories_rule.set_attribute_modifier_function(
'href', lambda x: self.get_href(x, base_url))
parser.add_rule(categories_rule)
video_rule = ParserRule()
video_rule.add_activate_rule_level([('body', '', '')])
video_rule.add_process_rule_level('script', {})
video_rule.set_attribute_filter_function(
'data', lambda text: 'angular.' in text)
parser.add_rule(video_rule)
#
gallery_href_rule = ParserRule()
gallery_href_rule.add_activate_rule_level([('div', 'class',
'row tag-area')])
gallery_href_rule.add_process_rule_level('a', {'href'})
gallery_href_rule.set_attribute_modifier_function(
'href', lambda x: self.get_href(x, base_url))
parser.add_rule(gallery_href_rule)
self.proceed_parcing(parser, fname)
result = ParseResult()
if video_rule.is_result(): # len(video_rule.get_result()) > 0:
script = video_rule.get_result()[0]['data'].replace(' ', '')
json_file_url = self.get_href(self.quotes(script, "host:'", "'"),
base_url)
# print(json_file_url)
from requests_loader import load, LoaderError
json_file = Setting.base_dir + 'tsp_video.json'
urls = list()
result.set_type('video')
try:
r = load(URL(json_file_url), json_file)
links = set()
for item in r.json()['mediaSources']:
# print(item)
if item['source'] not in links:
data = dict(text=item['quality'],
url=URL(item['source'] + '*'))
urls.append(data)
links.add(item['source'])
if len(urls) == 1:
video = MediaData(urls[0]['url'])
elif len(urls) > 1:
video = MediaData(urls[0]['url'])
for item in urls:
video.add_alternate(item)
else:
return result
result.set_video(video)
except LoaderError as err:
print(err)
for f in gallery_href_rule.get_result(['data', 'href']):
result.add_control(
ControlInfo(f['data'].strip(), URL(f['href'])))
return result
if startpage_rule.is_result(): # len(startpage_rule.get_result()) > 0:
result.set_type('hrefs')
for item in startpage_rule.get_result(['href']):
# print(item)
result.add_thumb(
ThumbInfo(thumb_url=URL(item['style']),
href=URL(item['href']),
popup=item.get('alt', '')))
for item in startpage_pages_rule.get_result(['href', 'data']):
label = item['data'].replace(' ', '')
# print(item)
if len(label) > 0:
result.add_page(ControlInfo(label, URL(item['href'])))
if categories_rule.is_result(['href']):
for item in categories_rule.get_result(['href', 'data']):
result.add_control(
ControlInfo(item['data'], URL(item['href'])))
if tags_rule.is_result(['href']):
for item in tags_rule.get_result(['href', 'data']):
result.add_control(
ControlInfo(item['data'], URL(item['href'])))
return result
def parse_index_file(self, fname, base_url=URL()):
parser = SiteParser()
startpage_rule = ParserRule()
startpage_rule.add_activate_rule_level([('div', 'class', 'inner_wrap')
])
startpage_rule.add_process_rule_level('a', {'href'})
startpage_rule.add_process_rule_level('img',
{'src', 'alt', 'data-original'})
# startpage_rule.set_attribute_modifier_function('href', lambda x: self.get_href(x,base_url) + '*' )
parser.add_rule(startpage_rule)
startpage_pages_rule = ParserRule()
startpage_pages_rule.add_activate_rule_level([('div', 'class',
'pagination')])
# startpage_pages_rule.add_activate_rule_level([('a', 'class', 'current')])
startpage_pages_rule.add_process_rule_level('a', {'href'})
startpage_pages_rule.set_attribute_modifier_function(
'href', lambda x: base_url.domain() + x)
parser.add_rule(startpage_pages_rule)
startpage_hrefs_rule = ParserRule()
startpage_hrefs_rule.add_activate_rule_level([('div', 'class',
'holder_list')])
# startpage_hrefs_rule.add_activate_rule_level([('a', 'class', 'current')])
startpage_hrefs_rule.add_process_rule_level('a', {'href', 'title'})
# startpage_hrefs_rule.set_attribute_modifier_function('href', lambda x: self.get_href(x,base_url) + '*')
parser.add_rule(startpage_hrefs_rule)
#
video_rule = ParserRule()
video_rule.add_activate_rule_level([('div', 'class', 'player')])
video_rule.add_process_rule_level('script', {})
video_rule.set_attribute_filter_function(
'data', lambda text: 'video_url:' in text or 'jwplayer(' in text)
parser.add_rule(video_rule)
#
gallery_user_rule = ParserRule()
gallery_user_rule.add_activate_rule_level([('div', 'class', 'descrow')
])
gallery_user_rule.add_process_rule_level('a', {'href', 'title'})
gallery_user_rule.set_attribute_modifier_function(
'href', lambda x: x + 'public_videos/')
gallery_user_rule.set_attribute_filter_function(
'href', lambda x: '/members/' in x)
parser.add_rule(gallery_user_rule)
gallery_href_rule = ParserRule()
gallery_href_rule.add_activate_rule_level([('div', 'class', 'descrow')
])
gallery_href_rule.add_process_rule_level('a', {'href'})
# gallery_href_rule.set_attribute_modifier_function('href', lambda x: base_url.domain() + x + '*')
gallery_href_rule.set_attribute_filter_function(
'href', lambda x: '/categories/' in x or '/tags/' in x)
parser.add_rule(gallery_href_rule)
self.proceed_parcing(parser, fname)
result = ParseResult()
if video_rule.is_result(): # len(video_rule.get_result()) > 0:
script = video_rule.get_result()[0]['data'].replace(
' ', '') # .replace('\\','')
# print(script)
file = 'Not found!!!!'
if 'jwplayer(' in script:
file = script.partition('file:"')[2].partition('",')[0] # +'*'
elif "video_url:'" in script:
file = script.partition("video_url:'")[2].partition("',")[
0] # +'*'
# print(file)
video = MediaData(URL(file))
result.set_type('video')
result.set_video(video)
#
# for f in gallery_channel_rule.get_result(['data', 'href']):
# result.add_control(ControlInfo(f['data'], URL(f['href'])))
if gallery_user_rule.is_result():
f = gallery_user_rule.get_result(['href'])[0]
print(f)
result.add_control(
ControlInfo('"' + f['data'] + '"', URL(f['href'])))
for f in gallery_href_rule.get_result(['href']):
# print(f)
result.add_control(ControlInfo(f['data'], URL(f['href'])))
return result
if startpage_rule.is_result(): # len(startpage_rule.get_result()) > 0:
result.set_type('hrefs')
for item in startpage_rule.get_result(['href']):
# print(item)
result.add_thumb(
ThumbInfo(thumb_url=URL(item['data-original']),
href=URL(item['href']),
popup=item.get('alt', '')))
for item in startpage_pages_rule.get_result(['href', 'data']):
result.add_page(ControlInfo(item['data'], URL(item['href'])))
if len(startpage_hrefs_rule.get_result(['href'])) > 0:
for item in startpage_hrefs_rule.get_result(['href', 'data']):
result.add_control(
ControlInfo(item.get('title', item.get('data', '')),
URL(item['href'])))
return result
def parse_index_file(self, fname, base_url=URL()):
parser = SiteParser()
startpage_rule = ParserRule()
startpage_rule.add_activate_rule_level([
('div', 'class', 'well well-sm hover'),
('div', 'class', 'channelContainer')
])
startpage_rule.add_process_rule_level('a', {'href', 'title'})
startpage_rule.add_process_rule_level('img', {'src', 'alt'})
startpage_rule.set_attribute_modifier_function(
'href', lambda x: self.get_href(x, base_url))
startpage_rule.set_attribute_modifier_function(
'src', lambda x: self.get_href(x, base_url))
parser.add_rule(startpage_rule)
startpage_pages_rule = ParserRule()
startpage_pages_rule.add_activate_rule_level([('ul', 'class',
'pagination')])
# startpage_pages_rule.add_activate_rule_level([('a', 'class', 'current')])
startpage_pages_rule.add_process_rule_level('a', {'href'})
startpage_pages_rule.set_attribute_modifier_function(
'href', lambda x: self.get_href(x, base_url))
parser.add_rule(startpage_pages_rule)
startpage_hrefs_rule = ParserRule()
startpage_hrefs_rule.add_activate_rule_level([('ul', 'class',
'drop2 hidden-xs')])
# startpage_hrefs_rule.add_activate_rule_level([('a', 'class', 'current')])
startpage_hrefs_rule.add_process_rule_level('a', {'href'})
# startpage_hrefs_rule.set_attribute_filter_function('href',lambda x: '/videos/' in x)
startpage_hrefs_rule.set_attribute_modifier_function(
'href', lambda x: self.get_href(x, base_url))
parser.add_rule(startpage_hrefs_rule)
#
video_rule = ParserRule()
video_rule.add_activate_rule_level([('div', 'class', 'video-container')
])
video_rule.add_process_rule_level('source', {'src', 'label', 'res'})
# video_rule.set_attribute_filter_function('data', lambda text: 'video_url' in text)
video_rule.set_attribute_modifier_function(
'src', lambda x: self.get_href(x, base_url))
parser.add_rule(video_rule)
#
gallery_href_rule = ParserRule()
gallery_href_rule.add_activate_rule_level([('div', 'class',
'm-t-10 overflow-hidden')])
gallery_href_rule.add_process_rule_level('a', {'href'})
gallery_href_rule.set_attribute_modifier_function(
'href', lambda x: self.get_href(x, base_url))
parser.add_rule(gallery_href_rule)
gallery_user_rule = ParserRule(collect_data=True)
gallery_user_rule.add_activate_rule_level([
('div', 'class', 'pull-left user-container')
])
gallery_user_rule.add_process_rule_level('a', {'href'})
gallery_user_rule.set_attribute_modifier_function(
'href', lambda x: self.get_href(x, base_url))
gallery_user_rule.set_attribute_filter_function(
'href', lambda x: '#' not in x)
parser.add_rule(gallery_user_rule)
self.proceed_parcing(parser, fname)
result = ParseResult()
if video_rule.is_result():
urls = UrlList()
for item in video_rule.get_result():
urls.add(item['res'], URL(item['src']))
result.set_video(urls.get_media_data(-1))
for f in gallery_user_rule.get_result(['data', 'href']):
result.add_control(
ControlInfo('"' + f['data'].strip() + '"', URL(f['href'])))
for f in gallery_href_rule.get_result(['data', 'href']):
result.add_control(ControlInfo(f['data'], URL(f['href'])))
return result
if startpage_rule.is_result():
for item in startpage_rule.get_result(['href']):
result.add_thumb(
ThumbInfo(thumb_url=URL(item['src']),
href=URL(item['href']),
popup=item.get('alt', item.get('title', ''))))
for item in startpage_pages_rule.get_result(['href', 'data']):
result.add_page(ControlInfo(item['data'], URL(item['href'])))
for item in startpage_hrefs_rule.get_result(['href', 'data']):
label = item['href'].strip('*').rpartition('/')[2]
result.add_control(ControlInfo(label, URL(item['href'])))
return result
def parse_index_file(self, fname, base_url=URL()):
parser = SiteParser()
startpage_rule = ParserRule()
startpage_rule.add_activate_rule_level([('div', 'class', 'mb'),
('div', 'class', 'mbhd'),
('div', 'class', 'mb mbr'),
('div', 'class', 'mbhd mbr'),
('div', 'class', 'categoriesbox')])
startpage_rule.add_process_rule_level('a', {'href'})
startpage_rule.add_process_rule_level('img', {'src', 'alt'})
startpage_rule.set_attribute_modifier_function('href', lambda x: base_url.domain() + x)
parser.add_rule(startpage_rule)
startpage_pages_rule = ParserRule()
startpage_pages_rule.add_activate_rule_level([('div', 'class', 'numlist2')])
startpage_pages_rule.add_process_rule_level('a', {'href'})
startpage_pages_rule.set_attribute_modifier_function('href', lambda x: base_url.domain() + x)
parser.add_rule(startpage_pages_rule)
video_rule = ParserRule()
video_rule.add_activate_rule_level([('div', 'id', 'moviexxx')])
video_rule.add_process_rule_level('script', {})
video_rule.set_attribute_filter_function('data', lambda x: "videojs('EPvideo'" in x)
parser.add_rule(video_rule)
video_fname_rule = ParserRule()
video_fname_rule.add_activate_rule_level([('div', 'id', 'hd-p**n-dload')])
video_fname_rule.add_process_rule_level('a', {'href'})
# video_fname_rule.set_attribute_filter_function('data',lambda x:"videojs('EPvideo'" in x)
parser.add_rule(video_fname_rule)
gallery_href_rule = ParserRule()
gallery_href_rule.add_activate_rule_level([('div', 'class', 'tab-1')])
# gallery_href_rule.add_activate_rule_level([('td', 'class', 'btnList')])
gallery_href_rule.add_process_rule_level('a', {'href'})
gallery_href_rule.set_attribute_modifier_function('href', lambda x: base_url.domain() + x)
gallery_href_rule.set_attribute_filter_function('href', lambda x: '/category/' in x or '/search/' in x)
parser.add_rule(gallery_href_rule)
self.proceed_parcing(parser, fname)
result = ParseResult()
if video_rule.is_result():
script = video_rule.get_result()[0]['data'].replace(' ', '')
# print(script)
vid = script.partition("vid:'")[2].partition("',")[0]
hash = script.partition("hash:'")[2].partition("',")[0]
print(vid, hash)
for item in video_fname_rule.get_result():
print(item)
fname = item['href'].rpartition('/')[2]
print(fname)
# print('http://v1.s1.n10.nl.cdn.eporner.com/3e7475b4b977ffd9649a87353717fc51/582c90c9027c00/310073-720p.mp4'.format(hash,vid,fname))
print('http://v1.s1.n10.nl.cdn.eporner.com/{0}/{1}/{2}'.format(hash, vid, fname))
# script_url=URL(base_url.domain() + (video_rule.get_result()[0]['data'].partition("getScript('")[2].partition("'")[0]))
# video=EPvideoParser(script_url,self.model)
# result.set_video(video.get_result())
result.set_type('video')
for f in gallery_href_rule.get_result(['data', 'href']):
# print(f)
result.add_control(ControlInfo(f['data'], URL(f['href'])))
# print('return')
return result
if len(startpage_rule.get_result()) > 0:
result.set_type('hrefs')
for item in startpage_rule.get_result():
result.add_thumb(ThumbInfo(thumb_url=URL(item['src']), href=URL(item['href']), popup=item['alt']))
for item in startpage_pages_rule.get_result(['href', 'data']):
result.add_page(ControlInfo(item['data'], URL(item['href'])))
#
# for item in startpage_hrefs_rule.get_result(['href', 'data']):
# result.add_control(ControlInfo(item['data'], URL(item['href'])))
return result
def parse_index_file(self, fname, base_url=URL()):
parser = SiteParser()
startpage_rule = ParserRule()
startpage_rule.add_activate_rule_level([('div', 'class', 'link-3col')])
startpage_rule.add_process_rule_level('a', {'href'})
startpage_rule.add_process_rule_level('img', {'src', 'alt'})
startpage_rule.set_attribute_modifier_function(
'href', lambda x: self.get_href(x, base_url))
startpage_rule.set_attribute_modifier_function(
'src', lambda x: self.get_href(x, base_url))
parser.add_rule(startpage_rule)
startpage_pages_rule = ParserRule()
startpage_pages_rule.add_activate_rule_level([('div', 'class',
'wp-pagenavi')])
startpage_pages_rule.add_process_rule_level('a', {'href'})
startpage_pages_rule.set_attribute_modifier_function(
'href', lambda x: self.get_href(x, base_url))
parser.add_rule(startpage_pages_rule)
tags_rule = ParserRule()
tags_rule.add_activate_rule_level([('div', 'class', 'tagcloud')])
tags_rule.add_process_rule_level('a', {'href'})
tags_rule.set_attribute_modifier_function(
'href', lambda x: self.get_href(x, base_url))
parser.add_rule(tags_rule)
video_rule = ParserRule()
video_rule.add_activate_rule_level([('div', 'class', 'player')])
video_rule.add_process_rule_level('iframe', {'src'})
video_rule.set_attribute_filter_function('src',
lambda x: 'fileone.tv' in x)
video_rule.set_attribute_modifier_function(
'src', lambda x: self.get_href(x, base_url))
parser.add_rule(video_rule)
gallery_href_rule = ParserRule()
gallery_href_rule.add_activate_rule_level([('div', 'id', 'extras')])
gallery_href_rule.add_process_rule_level('a', {'href'})
gallery_href_rule.set_attribute_modifier_function(
'href', lambda x: self.get_href(x, base_url))
parser.add_rule(gallery_href_rule)
self.proceed_parcing(parser, fname)
result = ParseResult()
if video_rule.is_result():
urls = UrlList()
for item in video_rule.get_result():
print(item)
try:
r = load(URL(item['src']))
setup = self.quotes(r.text, "jwplayer('player').setup(",
")").replace(' ', '')
file = self.quotes(setup, "file:'", "'")
urls.add("default", URL(file + '*'))
except LoaderError as err:
print(err)
result.set_video(urls.get_media_data())
for f in gallery_href_rule.get_result(['data', 'href']):
result.add_control(
ControlInfo(f['data'].strip(), URL(f['href'])))
return result
if startpage_rule.is_result(): # len(startpage_rule.get_result()) > 0:
for item in startpage_rule.get_result(['href']):
result.add_thumb(
ThumbInfo(thumb_url=URL(item['src']),
href=URL(item['href']),
popup=item.get('alt', '')))
for item in startpage_pages_rule.get_result(['href', 'data']):
result.add_page(ControlInfo(item['data'], URL(item['href'])))
for item in tags_rule.get_result(['href', 'data']):
result.add_control(ControlInfo(item['data'],
URL(item['href'])))
return result
def parse_index_file(self, fname, base_url=URL()):
parser = SiteParser()
startpage_rule = ParserRule()
startpage_rule.add_activate_rule_level([('div', 'class',
'contents videos')])
startpage_rule.add_process_rule_level('a', {'href', 'title'})
startpage_rule.add_process_rule_level('img', {'src', 'alt'})
startpage_rule.set_attribute_modifier_function(
'href', lambda x: self.get_href(x, base_url))
startpage_rule.set_attribute_modifier_function(
'src', lambda x: self.get_href(x, base_url))
parser.add_rule(startpage_rule)
startpage_pages_rule = ParserRule()
startpage_pages_rule.add_activate_rule_level([('div', 'id',
'pagination')])
# startpage_pages_rule.add_activate_rule_level([('a', 'class', 'current')])
startpage_pages_rule.add_process_rule_level('a', {'href'})
startpage_pages_rule.set_attribute_modifier_function(
'href', lambda x: self.get_href(x, base_url))
parser.add_rule(startpage_pages_rule)
#
video_rule = ParserRule()
video_rule.add_activate_rule_level([('div', 'class', 'contents')])
video_rule.add_process_rule_level('script', {})
video_rule.set_attribute_filter_function(
'data', lambda text: 'jwplayer' in text)
# video_rule.set_attribute_modifier_function('src',lambda x:self.get_href(x,base_url))
parser.add_rule(video_rule)
#
gallery_href_rule = ParserRule()
gallery_href_rule.add_activate_rule_level([('div', 'class',
'info_holder')])
gallery_href_rule.add_activate_rule_level([('div', 'class', 'l')])
gallery_href_rule.add_process_rule_level('a', {'href'})
gallery_href_rule.set_attribute_filter_function(
'href', lambda x: '/profiles/' not in x)
gallery_href_rule.set_attribute_modifier_function(
'href', lambda x: self.get_href(x, base_url))
parser.add_rule(gallery_href_rule)
gallery_user_rule = ParserRule()
gallery_user_rule.add_activate_rule_level([('div', 'class',
'info_holder')])
gallery_user_rule.add_process_rule_level('a', {'href'})
gallery_user_rule.set_attribute_filter_function(
'href', lambda x: '/profiles/' in x)
gallery_user_rule.set_attribute_modifier_function(
'href',
lambda x: self.get_href(x.replace('.html', '/videos/'), base_url))
parser.add_rule(gallery_user_rule)
self.proceed_parcing(parser, fname)
result = ParseResult()
if video_rule.is_result():
urls = UrlList()
for item in video_rule.get_result():
file = self.quotes(item['data'], 'file:', ',').strip(' "')
urls.add('default', URL(file + '*'))
result.set_video(urls.get_media_data())
for f in gallery_user_rule.get_result(['data', 'href']):
result.add_control(
ControlInfo('"' + f['data'].strip() + '"', URL(f['href'])))
for f in gallery_href_rule.get_result(['data', 'href']):
result.add_control(ControlInfo(f['data'], URL(f['href'])))
return result
if startpage_rule.is_result():
for item in startpage_rule.get_result(['href']):
result.add_thumb(
ThumbInfo(thumb_url=URL(item['src']),
href=URL(item['href']),
popup=item.get('alt', item.get('title', ''))))
for item in startpage_pages_rule.get_result(['href', 'data']):
result.add_page(ControlInfo(item['data'], URL(item['href'])))
return result
def parse_index_file(self, fname, base_url=URL()):
parser = SiteParser()
# print(base_url.domain())
def star_get_url(txt=''):
return txt.partition('(')[2].partition(')')[0]
startpage_rule = ParserRule(debug=False)
startpage_rule.add_activate_rule_level([('ul', 'class', 'video-listing'),
('ul', 'class', 'video-listing two-in-row'),
('ul', 'class', 'video-listing four-in-row'),
('ul', 'class', 'video-listing two-in-row id-recommended-list'),
('ul', 'class', 'video-listing four-in-row id-recommended-list')
])
startpage_rule.add_process_rule_level('a', {'href', 'title'})
startpage_rule.add_process_rule_level('img', {'data-src', 'src', 'alt'})
startpage_rule.set_attribute_modifier_function('href', lambda x: self.get_href(x, base_url))
startpage_rule.set_attribute_modifier_function('data-src', lambda x: self.get_href(x, base_url))
startpage_rule.set_attribute_modifier_function('src', lambda x: self.get_href(x, base_url))
parser.add_rule(startpage_rule)
startpage_pages_rule = ParserRule()
startpage_pages_rule.add_activate_rule_level([('div', 'class', 'pageNumbersHolder')])
# startpage_pages_rule.add_activate_rule_level([('a', 'class', 'current')])
startpage_pages_rule.add_process_rule_level('a', {'href'})
startpage_pages_rule.set_attribute_modifier_function('href', lambda x: self.get_href(x, base_url))
parser.add_rule(startpage_pages_rule)
startpage_hrefs_rule = ParserRule()
startpage_hrefs_rule.add_activate_rule_level([('ul', 'class', 'categories-listing'),
('ul', 'class', 'categories-popular-listing'),
('ul', 'class', 'abc-categories newAbcCategories')])
# startpage_hrefs_rule.add_activate_rule_level([('a', 'class', 'current')])
startpage_hrefs_rule.add_process_rule_level('a', {'href', 'title'})
startpage_hrefs_rule.set_attribute_modifier_function('href', lambda x: self.get_href(x, base_url))
parser.add_rule(startpage_hrefs_rule)
pornstars_rule = ParserRule()
pornstars_rule.add_activate_rule_level([('div', 'id', 'all_pornstars')])
pornstars_rule.add_process_rule_level('a', {'href'})
pornstars_rule.add_process_rule_level('img', {'src', 'alt'})
# pornstars_rule.set_attribute_filter_function('href',lambda x: '/channel/' in x or '/prime/' in x)
pornstars_rule.set_attribute_modifier_function('href', lambda x: self.get_href(x, base_url))
parser.add_rule(pornstars_rule)
pornstars_hrefs_rule = ParserRule()
pornstars_hrefs_rule.add_activate_rule_level([('ul', 'class', 'abc-categories newAbcCategories')])
# startpage_hrefs_rule.add_activate_rule_level([('a', 'class', 'current')])
pornstars_hrefs_rule.add_process_rule_level('a', {'href'})
pornstars_hrefs_rule.set_attribute_modifier_function('href', lambda x: self.get_href(x, base_url))
parser.add_rule(pornstars_hrefs_rule)
channels_rule = ParserRule(debug=False)
channels_rule.add_activate_rule_level([('ul', 'class', 'channels-list three-in-row')])
channels_rule.add_process_rule_level('a', {'href'})
channels_rule.add_process_rule_level('img', {'src', 'data-src', 'alt'})
channels_rule.set_attribute_modifier_function('href', lambda x: self.get_href(x, base_url))
channels_rule.set_attribute_modifier_function('src', lambda x: self.get_href(x, base_url))
channels_rule.set_attribute_modifier_function('data-src', lambda x: self.get_href(x, base_url))
parser.add_rule(channels_rule)
channels_hrefs_rule = ParserRule()
channels_hrefs_rule.add_activate_rule_level([('div', 'class', 'channel-filters-categories')])
# channels_hrefs_rule.add_activate_rule_level([('a', 'class', 'current')])
channels_hrefs_rule.add_process_rule_level('a', {'href', 'title'})
channels_hrefs_rule.set_attribute_modifier_function('href', lambda x: self.get_href(x, base_url))
parser.add_rule(channels_hrefs_rule)
video_rule = ParserRule()
video_rule.add_activate_rule_level([('div', 'class', 'watch')])
video_rule.add_process_rule_level('script', {})
video_rule.set_attribute_filter_function('data', lambda text: 'redtube_flv_player' in text)
parser.add_rule(video_rule)
#
gallery_href_rule = ParserRule()
gallery_href_rule.add_activate_rule_level([('div', 'class', 'video-details')])
# gallery_href_rule.add_activate_rule_level([('td', 'class', 'links')])
gallery_href_rule.add_process_rule_level('a', {'href'})
gallery_href_rule.set_attribute_modifier_function('href', lambda x: self.get_href(x, base_url))
gallery_href_rule.set_attribute_filter_function('href', lambda x: x != '*')
parser.add_rule(gallery_href_rule)
self.proceed_parcing(parser, fname)
result = ParseResult()
if video_rule.is_result():
urls = UrlList()
for item in video_rule.get_result():
script = item['data'].replace(' ', '').replace('\\', '')
sources = self.quotes(script, 'sources:{"', '"},').split('","')
for f in sources:
t = f.partition('":"')
label = t[0]
file = self.get_href(t[2], base_url)
urls.add(label, URL(file))
result.set_video(urls.get_media_data())
for f in gallery_href_rule.get_result(['data', 'href']):
href = f['href']
label = f['data']
if '/redtube/' in href or '/tag/' in href:
result.add_control(ControlInfo(label, URL(href)))
elif '/pornstar/' in href:
ps_name = href.rstrip('*').rpartition('/')[2].replace('+', ' ').title()
result.add_control(ControlInfo(ps_name, URL(href)))
else:
# adding user
result.add_control(ControlInfo("'" + label + "'", URL(href.replace('*', '/videos*'))))
return result
if pornstars_rule.is_result():
result.set_caption_visible(True)
for item in pornstars_rule.get_result():
result.add_thumb(
ThumbInfo(thumb_url=URL(item['src']), href=URL(item['href']), popup=item.get('alt', '')))
for item in startpage_pages_rule.get_result(['href', 'data']):
result.add_page(ControlInfo(item['data'], URL(item['href'])))
for item in pornstars_hrefs_rule.get_result(['href']):
result.add_control(ControlInfo(item['data'], URL(item['href'])))
return result
if channels_rule.is_result():
result.set_caption_visible(True)
for item in channels_rule.get_result(['href']):
thumb_href = item.get('data-src', item.get('src'))
descr = item.get('alt', '').title()
result.add_thumb(ThumbInfo(thumb_url=URL(thumb_href), href=URL(item['href']), popup=descr))
for item in startpage_pages_rule.get_result(['href', 'data']):
result.add_page(ControlInfo(item['data'], URL(item['href'])))
for item in channels_hrefs_rule.get_result(['href']):
result.add_control(ControlInfo(item['title'], URL(item['href'])))
return result
if startpage_rule.is_result():
for item in startpage_rule.get_result(['href']):
thumb_href = item.get('data-src', item.get('src'))
descr = item.get('title', item.get('alt', ''))
result.add_thumb(ThumbInfo(thumb_url=URL(thumb_href), href=URL(item['href']), popup=descr))
for item in startpage_pages_rule.get_result(['href', 'data']):
result.add_page(ControlInfo(item['data'], URL(item['href'])))
for item in startpage_hrefs_rule.get_result(['href', 'title']):
result.add_control(ControlInfo(item['title'], URL(item['href'])))
return result
def parse_index_file(self, fname, base_url=URL()):
print(base_url.domain())
parser = SiteParser()
startpage_rule = ParserRule()
startpage_rule.add_activate_rule_level([('div', 'class', 'ownpost')])
startpage_rule.add_process_rule_level('a', {'href'})
startpage_rule.add_process_rule_level('img', {'src', 'alt'})
startpage_rule.set_attribute_modifier_function(
'href', lambda x: self.get_href(x, base_url))
parser.add_rule(startpage_rule)
startpage_pages_rule = ParserRule()
startpage_pages_rule.add_activate_rule_level([('div', 'class', 'nav')])
startpage_pages_rule.add_process_rule_level('a', {'href'})
startpage_pages_rule.set_attribute_modifier_function(
'href', lambda x: base_url.domain() + '/' + x + '*')
parser.add_rule(startpage_pages_rule)
picture_rule = ParserRule()
picture_rule.add_activate_rule_level([('div', 'class', 'gallery')])
picture_rule.add_process_rule_level('a', set())
picture_rule.add_process_rule_level('img', {'src', 'class'})
picture_rule.set_attribute_modifier_function(
'src', lambda text: text.replace('/tn_', '/'))
parser.add_rule(picture_rule)
picture_href_rule = ParserRule()
picture_href_rule.add_activate_rule_level([('div', 'class', 'gallery')
])
picture_href_rule.add_process_rule_level('a', {'href'})
picture_href_rule.set_attribute_modifier_function(
'href', lambda x: self.get_href(x, base_url))
picture_href_rule.set_attribute_filter_function(
'href', lambda x: x.find('/?category=') != -1)
parser.add_rule(picture_href_rule)
for s in open(fname):
parser.feed(s)
result = ParseResult()
if len(startpage_rule.get_result()) > 0:
# print('Startpage rule')
result.set_type('hrefs')
for item in startpage_rule.get_result():
result.add_thumb(
ThumbInfo(thumb_url=URL(item['src']),
href=URL(item['href'] + '*'),
popup=item.get('alt', '')))
for item in startpage_pages_rule.get_result(['href', 'data']):
result.add_page(ControlInfo(item['data'], URL(item['href'])))
if len(picture_rule.get_result()) > 0:
result.set_type('pictures')
i = 1
for f in picture_rule.get_result(['src', 'class']):
# print(f)
result.add_full(
FullPictureInfo(abs_href=URL(f['src']),
rel_name='%03d.jpg' % i))
i += 1
for f in picture_href_rule.get_result():
# print(f)
result.add_control(
ControlInfo(f['data'].replace(',', ''), URL(f['href'])))
return result
def parse_index_file(self, fname, base_url=URL()):
print('VP parsing')
parser = SiteParser()
startpage_rule = ParserRule()
startpage_rule.add_activate_rule_level([('div', 'class', 'bx'),
('div', 'class', 'bx lastrow')
])
startpage_rule.add_process_rule_level('a', {'href', 'class'})
startpage_rule.add_process_rule_level('img', {'src', 'alt'})
parser.add_rule(startpage_rule)
startpage_pages_rule = ParserRule()
startpage_pages_rule.add_activate_rule_level([('div', 'class',
'pagerwrap')])
# startpage_pages_rule.add_activate_rule_level([('a', 'class', 'current')])
startpage_pages_rule.add_process_rule_level('a', {'href'})
# startpage_pages_rule.set_attribute_modifier_function('href',lambda x: base_url.domain() + x)
parser.add_rule(startpage_pages_rule)
video_rule = ParserRule()
video_rule.add_activate_rule_level([('div', 'class', 'video_panel')])
video_rule.add_process_rule_level('script', {})
video_rule.set_attribute_filter_function(
'data', lambda text: 'var flashvars' in text)
parser.add_rule(video_rule)
gallery_href_rule = ParserRule()
gallery_href_rule.add_activate_rule_level([('div', 'class',
'tagas-secondrow')])
# gallery_href_rule.add_activate_rule_level([('td', 'class', 'btnList')])
gallery_href_rule.add_process_rule_level('a', {'href'})
gallery_href_rule.set_attribute_modifier_function(
'href', lambda x: base_url.domain() + x + '*')
# gallery_href_rule.set_attribute_filter_function('href',lambda x:'/category/'in x or '/search/'in x)
parser.add_rule(gallery_href_rule)
gallery_user_rule = ParserRule()
gallery_user_rule.add_activate_rule_level([('div', 'class', 'info')])
gallery_user_rule.add_process_rule_level('a', {'href'})
# gallery_user_rule.set_attribute_filter_function('href',lambda x:'/profile/' in x)
gallery_user_rule.set_attribute_modifier_function(
'href', lambda x: self.get_href(x.replace('/user/', '/submitted/'),
base_url))
parser.add_rule(gallery_user_rule)
for s in open(fname, encoding='utf-8', errors='ignore'):
# print(s)
parser.feed(s.replace('</b>', '</a>'))
result = ParseResult()
if len(video_rule.get_result()) > 0:
script = video_rule.get_result()[0]['data'].replace(' ', '')
# print(script)
def get_url_from_script(script='', var=''):
data = script.partition('flashvars.' + var +
'="')[2].partition('"')[0]
# print(var,data)
if data.startswith('https://'): return URL(data)
videoUrlLow = get_url_from_script(script, 'videoUrlLow')
videoUrlLow2 = get_url_from_script(script, 'videoUrlLow2')
videoUrlMedium = get_url_from_script(script, 'videoUrlMedium')
videoUrlMedium2 = get_url_from_script(script, 'videoUrlMedium2')
videoUrlHD = get_url_from_script(script, 'videoUrlHD')
videoUrlHD2 = get_url_from_script(script, 'videoUrlHD2')
def add_alternate(video, txt, url):
if url is not None:
video.add_alternate(dict(text=txt, url=url))
# video=MediaData(videoUrlMedium)
if videoUrlMedium is not None:
video = MediaData(videoUrlMedium)
elif videoUrlLow is not None:
video = MediaData(videoUrlLow)
else:
print('No url found')
return result
add_alternate(video, 'Low', videoUrlLow)
add_alternate(video, 'Low2', videoUrlLow2)
add_alternate(video, 'Medium', videoUrlMedium)
add_alternate(video, 'Medium', videoUrlMedium2)
add_alternate(video, 'HD', videoUrlHD)
add_alternate(video, 'HD', videoUrlHD2)
result.set_type('video')
result.set_video(video)
for f in gallery_user_rule.get_result():
# print(f)
result.add_control(
ControlInfo('"' + f['data'] + '"', URL(f['href'])))
for f in gallery_href_rule.get_result(['data', 'href']):
# print(f)
result.add_control(ControlInfo(f['data'], URL(f['href'])))
# print('return')
return result
if len(startpage_rule.get_result()) > 0:
result.set_type('hrefs')
for item in startpage_rule.get_result():
result.add_thumb(
ThumbInfo(thumb_url=URL(item['src']),
href=URL(item['href']),
popup=item['alt']))
for item in startpage_pages_rule.get_result(['href', 'data']):
result.add_page(ControlInfo(item['data'], URL(item['href'])))
# for item in startpage_hrefs_rule.get_result(['href', 'data']):
# result.add_control(ControlInfo(item['data'], URL(item['href'])))
return result
def parse_index_file(self, fname, base_url=URL()):
parser = SiteParser()
startpage_rule = ParserRule()
startpage_rule.add_activate_rule_level([('div', 'class',
'content-inner')])
startpage_rule.add_process_rule_level('a', {'href'})
startpage_rule.add_process_rule_level('img', {'src', 'alt'})
startpage_rule.set_attribute_modifier_function(
'href', lambda x: self.get_href(x, base_url) + '*')
parser.add_rule(startpage_rule)
startpage_pages_rule = ParserRule()
startpage_pages_rule.add_activate_rule_level([('div', 'class',
'pagination_link')])
# startpage_pages_rule.add_activate_rule_level([('a', 'class', 'current')])
startpage_pages_rule.add_process_rule_level('a', {'href'})
startpage_pages_rule.set_attribute_modifier_function(
'href', lambda x: base_url.domain() + x + '*')
parser.add_rule(startpage_pages_rule)
startpage_hrefs_rule = ParserRule()
startpage_hrefs_rule.add_activate_rule_level([
('div', 'class', 'sub_menu dark-menu'),
('div', 'class', 'sub-menu dark-menu')
])
# startpage_hrefs_rule.add_activate_rule_level([('a', 'class', 'current')])
startpage_hrefs_rule.add_process_rule_level('a', {'href', 'title'})
startpage_hrefs_rule.set_attribute_modifier_function(
'href', lambda x: self.get_href(x, base_url) + '*')
parser.add_rule(startpage_hrefs_rule)
#
video_rule = ParserRule()
video_rule.add_activate_rule_level([('div', 'id', 'content')])
video_rule.add_process_rule_level('script', {})
video_rule.set_attribute_filter_function(
'data', lambda text: 'jwplayer' in text)
parser.add_rule(video_rule)
#
gallery_href_rule = ParserRule()
gallery_href_rule.add_activate_rule_level([('div', 'id',
'media-tags-container')])
gallery_href_rule.add_process_rule_level('a', {'href'})
gallery_href_rule.set_attribute_modifier_function(
'href', lambda x: base_url.domain() + x + '*')
parser.add_rule(gallery_href_rule)
gallery_user_rule = ParserRule()
gallery_user_rule.add_activate_rule_level([('div', 'class',
'thumb-member-username')])
gallery_user_rule.add_process_rule_level('a', {'href'})
# gallery_user_rule.set_attribute_modifier_function('href', lambda x: base_url.domain() + x + '*')
# gallery_user_rule.set_attribute_filter_function('href',lambda x:'/categories/' in x)
parser.add_rule(gallery_user_rule)
self.proceed_parcing(parser, fname)
result = ParseResult()
if video_rule.is_result():
urls = UrlList()
for item in video_rule.get_result():
file = self.quotes(item['data'].replace(' ', ''), '"file":"',
'"')
urls.add('default', URL(file + '*'))
result.set_video(urls.get_media_data())
if gallery_user_rule.is_result():
user = gallery_user_rule.get_result()[0]['href'].rpartition(
'/')[2]
result.add_control(
ControlInfo('"' + user + ' uploads"',
URL('http://motherless.com/u/' + user + '*')))
result.add_control(
ControlInfo(
'"' + user + ' gals"',
URL('http://motherless.com/galleries/member/' + user +
'*')))
for f in gallery_href_rule.get_result(['data', 'href']):
result.add_control(ControlInfo(f['data'], URL(f['href'])))
return result
if startpage_rule.is_result():
for item in startpage_rule.get_result(['href']):
result.add_thumb(
ThumbInfo(thumb_url=URL(item['src']),
href=URL(item['href']),
popup=item.get('alt', '')))
for item in startpage_pages_rule.get_result(['href', 'data']):
result.add_page(ControlInfo(item['data'], URL(item['href'])))
for item in startpage_hrefs_rule.get_result(['href', 'data']):
result.add_control(
ControlInfo(item.get('title', item.get('data', '')),
URL(item['href'])))
return result
def parse_index_file(self, fname, base_url=URL()):
parser = SiteParser()
startpage_rule = ParserRule()
startpage_rule.add_activate_rule_level([('div', 'class', 'block-video')
])
startpage_rule.add_process_rule_level('a', {'href'})
startpage_rule.add_process_rule_level('img', {'src'})
# startpage_rule.set_attribute_modifier_function('href', lambda x: base_url.domain() + x)
parser.add_rule(startpage_rule)
startpage_pages_rule = ParserRule()
startpage_pages_rule.add_activate_rule_level([('ul', 'class',
'pagination')])
startpage_pages_rule.add_process_rule_level('a', {'title', 'href'})
startpage_pages_rule.set_attribute_modifier_function(
'href', lambda x: base_url.domain() + x)
parser.add_rule(startpage_pages_rule)
startpage_hrefs_rule = ParserRule()
startpage_hrefs_rule.add_activate_rule_level([('li', 'id',
'categories')])
startpage_hrefs_rule.add_process_rule_level('a', {'href'})
# startpage_hrefs_rule.set_attribute_modifier_function('href', lambda x: base_url.domain() + x)
parser.add_rule(startpage_hrefs_rule)
video_rule = ParserRule()
video_rule.add_activate_rule_level([('div', 'class', 'player_body')])
video_rule.add_process_rule_level('script', {})
video_rule.set_attribute_filter_function(
'data', lambda text: 'function playStart()' in text)
parser.add_rule(video_rule)
picture_href_rule = ParserRule()
picture_href_rule.add_activate_rule_level([
('p', 'class', 'info_category'), ('p', 'class', 'info_tags'),
('p', 'class', 'info_cast')
])
picture_href_rule.add_process_rule_level('a', {'href'})
parser.add_rule(picture_href_rule)
self.proceed_parcing(parser, fname)
result = ParseResult()
if len(video_rule.get_result()) > 0:
result.set_video(
MediaData(
URL(
self.get_attr_from_script(
video_rule.get_result()[0]['data']))))
result.set_type('video')
for f in picture_href_rule.get_result():
# print(f)
result.add_control(ControlInfo(f['data'], URL(f['href'])))
return result
if len(startpage_rule.get_result()) > 0:
result.set_type('hrefs')
for item in startpage_rule.get_result():
result.add_thumb(
ThumbInfo(thumb_url=URL(item['src']),
href=URL(item['href'])))
for item in startpage_pages_rule.get_result(['href', 'data']):
if '?from=' in item['href']:
result.add_page(
ControlInfo(item['data'], URL(item['href'] + '*')))
else:
result.add_page(
ControlInfo(item['data'], URL(item['href'])))
for item in startpage_hrefs_rule.get_result(['href', 'data']):
result.add_control(ControlInfo(item['data'],
URL(item['href'])))
return result
def parse_index_file(self, fname, base_url=URL()):
parser = SiteParser()
startpage_rule = ParserRule()
# startpage_rule.add_activate_rule_level([('section', '', '')])
startpage_rule.add_activate_rule_level([
('article', 'class', 'teaser singleLink hasButtonRow'),
('article', 'class', 'activity video hasButtonFooter')
])
startpage_rule.add_process_rule_level('a', {'href'})
startpage_rule.add_process_rule_level('img',
{'src', 'data-lazysrc', 'alt'})
startpage_rule.set_attribute_modifier_function(
'href', lambda x: self.get_href(x, base_url))
startpage_rule.set_attribute_modifier_function(
'src', lambda x: self.get_href(x, base_url))
parser.add_rule(startpage_rule)
startpage_pages_rule = ParserRule()
startpage_pages_rule.add_activate_rule_level([
('nav', 'class', 'clearfix pagination bottom'),
('nav', 'class', 'range rangeCount-2 clearfix')
])
startpage_pages_rule.add_process_rule_level('a', {'href', 'data-href'})
startpage_pages_rule.set_attribute_modifier_function(
'href', lambda x: self.get_href(x, base_url))
startpage_pages_rule.set_attribute_modifier_function(
'data-href', lambda x: self.get_href(x, base_url))
parser.add_rule(startpage_pages_rule)
startpage_categories_rule = ParserRule()
startpage_categories_rule.add_activate_rule_level([
('select', 'id', 'input_selectCategories')
])
startpage_categories_rule.add_process_rule_level('option', {'value'})
startpage_categories_rule.set_attribute_modifier_function(
'value', lambda x: self.get_href(x, base_url))
parser.add_rule(startpage_categories_rule)
video_rule = ParserRule()
video_rule.add_activate_rule_level([('div', 'id', 'playerWrapper')])
video_rule.add_process_rule_level('script', {})
video_rule.set_attribute_filter_function(
'data', lambda text: 'sources:' in text)
parser.add_rule(video_rule)
#
gallery_href_rule = ParserRule()
gallery_href_rule.add_activate_rule_level([('dl', 'class', 'group')])
gallery_href_rule.add_process_rule_level('a', {'href'})
gallery_href_rule.set_attribute_modifier_function(
'href', lambda x: self.get_href(x, base_url))
parser.add_rule(gallery_href_rule)
gallery_user_rule = ParserRule()
gallery_user_rule.add_activate_rule_level([
('nav', 'class', 'profileNav clearfix buttonRow')
])
gallery_user_rule.add_process_rule_level('a', {'href'})
gallery_user_rule.set_attribute_modifier_function(
'href', lambda x: self.get_href(x, base_url))
gallery_user_rule.set_attribute_filter_function(
'href', lambda x: '#videos' in x)
parser.add_rule(gallery_user_rule)
self.proceed_parcing(parser, fname)
result = ParseResult()
if video_rule.is_result():
urls = UrlList()
for item in video_rule.get_result():
script = item['data'].replace(' ', '').replace('\\', '')
sources = self.quotes(script, 'sources:{"', '"},').split('","')
for f in sources:
t = f.partition('":"')
label = t[0]
file = self.get_href(t[2], base_url)
urls.add(label, URL(file))
result.set_video(urls.get_media_data())
# for f in gallery_user_rule.get_result():
# print(f)
# name='"{0}"'.format(f['href'].rpartition('/')[2].partition('#')[0])
# result.add_control(ControlInfo(name, URL(f['href'])))
for f in gallery_href_rule.get_result(['data', 'href']):
result.add_control(ControlInfo(f['data'], URL(f['href'])))
return result
if startpage_rule.is_result():
for item in startpage_rule.get_result(['href']):
# print(item)
result.add_thumb(
ThumbInfo(thumb_url=URL(
item.get('data-lazysrc', item['src'])),
href=URL(item['href']),
popup=item.get('alt'
'')))
for item in startpage_pages_rule.get_result(['href', 'data']):
# print(item)
href = item.get('data-href', item['href'])
# print(href)
result.add_page(
ControlInfo(href.rpartition('/')[2].strip('*'), URL(href)))
for item in startpage_categories_rule.get_result():
result.add_control(
ControlInfo(item['data'], URL(item['value'])))
return result
